We identified 54 unique studies describing NCNP use by SES across 12 countries at stage IV of the smoking epidemic, with a majority of studies located in the USA (35). Nineteen studies focused exclusively on e-cigarettes, and 28 exclusively on smokeless tobacco (including snus). Three studies reported on e-cigarettes and smokeless tobaco. We identified only four studies focused on NRT (Supplementary tables), and no eligible studies on other types of NCNP. All studies reported NCNP use by SES based on cross-sectional data, most often from surveys but in a few cases using baseline data from longitudinal studies. Indicators of SES included measures based on income, education, occupational group, composite indices and neighbourhood disadvantage. Measures of NCNP use included current use and ever use, with both sometimes used in the same study. Thirteen studies (five for e-cigarettes and eight for smokeless tobacco) had study populations ranging from 10-20 years old and were analysed separately from adult studies (those with populations 16 years and over). One study with an age range of 14-31 years and mean age of 19.5 was considered with the young people studies . Definitions of ‘current’ and ‘ever use’ were not always consistent, but ‘current use’ most often referred to daily use, regular use or any use within the last 30 days; while ‘ever use’ was defined variously from any use in the past 12 months to ever lifetime use.
Seventeen studies examined e-cigarette use by SES in adults, of which seven met our definition of ‘best evidence’. No study included details on the type of e-liquid used, i.e. whether or not this contained nicotine. For all adult e-cigarette studies that provided relevant data, combustible tobacco use was higher among low SES groups – thus confirming our first assumption. The data collection period for the best evidence studies was 2010-2014.
Current use among adults
Evidence of the potential equity impacts of current e-cigarette use was somewhat inconsistent, but generally pointed towards positive equity impacts (Figure 3, 3b). For instance, of four US studies using large population-based samples, two [30, 31] found potentially positive impacts for current use by income, but unclear or neutral impacts by education; while one  found the reverse (i.e. potentially positive impacts by education but unclear impacts by income). The fourth US study suggested unclear impacts . A fifth study (from the UK) found potentially negative impacts of current e-cigarette use by occupational group ; while a sixth study (based on a small sample of e-cigarette users in the US) showed unclear impacts .
Ever use among adults
Studies reporting on e-cigarette ever use among adults showed a more consistent pattern, with potentially negative and unclear findings dominating (Figure 3c). Two studies from the US  indicated potentially negative equity impacts by income, with one showing unclear impacts by education; another US study indicated unclear outcomes by education and income . Potentially negative impacts by both education and income were also found in a study among UK smokers .
Low quality studies, adults
Lower quality studies that were excluded from our ‘best evidence’ synthesis [38–47] showed broadly similar patterns to those illustrated above for both current and ever use of e-cigarettes, with many finding no clear pattern by SES.
Of the five studies examining e-cigarette use by SES in young people, none met our criteria for ‘best evidence’. Of three studies examining current use, two studies found potentially positive impacts (one from Finland examining use by education , and another from Ireland examining use by neighbourhood deprivation ) while a third study (from the US) found unclear impacts by education . For e-cigarette ever use, the first two studies showed unclear findings [48, 49], as did a fourth study from France . A fifth study (from Canada) found potentially positive impacts for ever use . Overall, the limited quality and quantity of the evidence means no conclusions can be drawn on the likely equity impact of e-cigarette use among young people.
Twenty-three studies examined smokeless tobacco use in adults by SES, of which 16 met our criteria for ‘best evidence’. For all but one of these studies, combustible tobacco use was higher among lower SES groups. Three studies reported on older data (1971-1990) [53–55], with the other 13 studies reporting data from the last 15 years (2005-2014). The majority (13) of studies came from the US, with two studies from Sweden and one from Norway. The ban on smokeless tobacco throughout the EU (except Sweden) explains the scarcity of European evidence on this form of NCNP.
Current use among adults
Studies from the USA and Scandinavia showed broadly similar patterns in the distribution of smokeless tobacco use by SES, with higher use among low SES groups suggesting a potentially positive equity impact (Figure 4, 4b). Studies of smokeless tobacco prevalence among large population-based samples came exclusively from the US; these indicated potentially positive impacts for current use by education, but mostly neutral or unclear impacts by income [31, 54, 56–63]. A study from Sweden , provided data on the association between smokeless tobacco use and SES (measured by education and income), but no prevalence data. Similar to the US evidence, this study indicated potentially positive impacts of smokeless tobacco use by education but unclear impacts by income. All studies found very low smokeless tobacco use among women compared with men, such that these findings effectively represent only male use of smokeless tobacco. Indeed, three of the 16 studies included only men in their study populations [54, 58, 62].
Ever-use among adults
A majority of studies looking at ever use - including four from the US and one from Norway - similarly suggested mostly positive equity impacts for smokeless tobacco use by education, and a mix of potentially positive, neutral or unclear impacts by income [53, 54, 65–67] (Figure 4c). Interestingly, the study from Norway  focussed solely on women and was able to generate reliable data due to the relatively high use of snus among this population in Norway. A sixth study examining ever-use of snus among older adults (40-60 years) in Sweden  found unclear equity impacts by education.
Low quality studies, adults
A further six US studies were not included in our ‘best evidence’ synthesis due to limited quality. Two of these also suggested positive impacts from smokeless tobacco use, [68, 69] with the remaining four showing neutral or unclear patterns in use by SES [38, 46, 70–72].
Of eight studies examining smokeless tobacco use in young people only two met our ‘best evidence’ criteria, both conducted with secondary school students in Norway. One suggested higher smokeless tobacco current use among students with lower educational trajectories, but no clear pattern in relation to family income ; while the other found no gradient in smokeless tobacco current use by education . Of the remaining six studies (of low quality or relevance), most showed unclear [75–78] or potentially positive equity impacts for current smokeless tobacco use [29, 79], although one suggested potentially negative equity impacts .
Nicotine Replacement Therapy
We found only four eligible studies of NRT use by SES, of which just two met our criteria for best evidence. Both studies drew on surveys of adult smokers. One, a UK study based on national survey data, indicated a neutral equity impact by occupational group among smokers who had used NRT in quit attempts over the previous year . The other study  used data from four waves of an Australian national survey, and reported NRT use for cessation in the previous year; this indicated potentially negative equity impacts by income, but unclear impacts by education. The two low quality studies included one from Canada  suggesting neutral equity impacts by education and income; and another from the US  suggesting potentially positive impacts by education.