Discomfort, primarily due to heat, was the most widely identified reason why mosquito net owners chose not to use a mosquito net on one or more nights in the 17 survey-based studies included in this review. The next most widely reported reason for not using a mosquito net in the survey-based studies was perceived low mosquito density; although this only accounted for 12.3% of all responses in the pooled data set (compared to 47.5% for discomfort) suggesting it was widely reported, but often at a relatively low frequency. Heat and perceived mosquito density were also consistently identified in the small number of studies presenting qualitative data, although perceived mosquito density more clearly emerged as the dominant reason for not using a mosquito net in these studies. In one qualitative study participants reported that the primary function of mosquito nets was to combat perceived mosquito nuisance (i.e. disruption of sleep) rather than malaria transmission . This finding has been reported elsewhere [24, 27] and indicates that the practical function of mosquito nets may differ from the intended function in some instances. The reported use of mosquito nets for fishing would be an extreme example of a problematic discrepancy between intended and practical mosquito net function . Utilizing a mosquito net to minimize sleep disturbance rather than malaria transmission is considerably less problematic as protection against malaria transmission is still conferred when the net is in use. Indeed, social marketing campaigns have even promoted ITNs as a means to minimize sleep disturbance . However, if this discrepancy results in seasonal or irregular use (as the evidence presented in this review indicates) then the benefit of mosquito net utilization may not be fully realized. If a primary motivation to use a mosquito net is perceived mosquito density, then it also stands to reason that in areas where mosquito density falls as a result of increased ITN coverage the continued motivation to use an ITN may decrease. In other words, the very effectiveness of the ITN may render further ITN use undesirable. If such a scenario were to eventuate, then any programmatic gains in terms of reduced malaria-related morbidity and mortality, as well as the possibility of future malaria elimination, could be potentially threatened. The relationship between mosquito net use and mosquito density may, therefore, warrant careful and ongoing investigation in areas experiencing an increase in ITN coverage.
If personal discomfort and, to a lesser extent, perceived mosquito density are the primary reasons for not using a mosquito net amongst mosquito net owners, then greater mosquito net use could potentially be achieved amongst this population via education or behaviour change communication (BCC) strategies as previously suggested [11, 18]. With respect to personal discomfort, however, education or BCC strategies would do nothing to change the physical properties of the mosquito net that cause the discomfort in the first place. Thus, modifications to mosquito nets or the mosquito net using environment that render the mosquito net more comfortable would usefully complement any educational or BCC campaign.
This review identified other reasons for not using a mosquito net that may also be better addressed via strategies other than (or in addition to) education or BCC. The pooled survey data indicated that social factors, such as sleeping elsewhere, or not sleeping at all, frequently result in mosquito net non-use. Technical factors related to mosquito net use (i.e. not being able to hang a mosquito net or finding it inconvenient to hang) and the temporary unavailability of a mosquito net (primarily due to someone else using it) were also reported in the survey and qualitative studies. Social obstacles to mosquito net use may be addressed by complementary mosquito control strategies. For example, if someone is active (i.e. not in bed) during night time hours then insect repellents could be made available and their use promoted; a malaria control strategy previously trialled with some success [41, 42]. Alternatively, if an individual is required to sleep somewhere other than their normal residence then additional 'travel' mosquito nets could be made available.
Additional mosquito nets in the house would usefully address the issue of non-use due to the temporary unavailability of a normally available net. Individuals who spend part of the night sleeping outdoors and part of the night sleeping indoors, a reason for 'partial' non-use identified in this review  and reported elsewhere , may also benefit from additional mosquito nets if they were able to hang them in their various sleeping areas. Nevertheless, hanging mosquito nets outdoors may continue to be problematic given current mosquito net designs and their reliance on external supporting structures. Thus, the development of 'outdoor' or 'stand alone' mosquito nets that require no external supports yet remain portable and user friendly would be beneficial. Increasing mosquito net use via a reduction in the technical difficulties associated with hanging or using a net may also best be achieved at the manufacturer level. Innovative design solutions could potentially resolve the reported difficulties in hanging a net in certain household structures (e.g. via the incorporation of internal or complementary supporting structures, compact sizes, or alternative 'clamping/tying' systems) or increase the ease with which a net may be utilized once hung. Where technical issues remain problematic consideration could even be given to the promotion of LLIN hammocks, blankets or curtains or even insecticide treated plastic sheeting (as a wall covering).
Perhaps the most important finding of this review pertains to the current state of the published research literature, which was limited at best. A basic descriptive analysis of the reported 'reasons for not using a mosquito net' data was not provided in eight out of the seventeen survey-based studies included in this review. This omission rendered it impossible to reliably interpret the relative importance of the reported findings in the respective studies. In the nine studies in which detailed descriptive analysis was presented, reliable interpretation of the data was often undermined by inadequate description of the study design. Examples included the frequent failure to report how data were obtained (e.g. in response to a structured checklist or open-ended question) or how many responses per participant were permissible. The omissions described above most probably reflect the fact that the focus of this review (reasons for not using a mosquito net as reported by mosquito net owners) was rarely a primary focus of the studies identified by the search methodology. Rather, the reviewed data were typically a relatively minor component of broader investigations of mosquito net use or malaria-related beliefs and practices. This in itself is a significant finding as it indicates an important area of investigation in the current environment of mass mosquito net distribution - why people who own/have available mosquito nets choose not to use them - has received minimal, dedicated, research attention.
In considering the recommendations made above it is also incumbent upon the reviewers to acknowledge that the survey-based data included in this review do not lend themselves readily to generalization and the pooled data must be interpreted with considerable caution. The respective surveys were variously conducted at different seasons or periods of high/low malaria transmission, a threat to cross-survey comparison and minimal data were obtained from non-African participants. In fact, the majority of reported data came from specific and diverse African sub-populations (e.g. pregnant women, caregivers of children under 5) and were often population specific (e.g. failure to use a mosquito net due to child-birth). Even seemingly generic reasons for not using a mosquito net, such as heat or perceived low mosquito density, may be more or less pertinent to specific populations, yet the quantity of currently available data is not sufficient to allow any such patterns to reliably emerge. Measures of mosquito net non-use were equally varied across the reviewed studies (rendering comparison difficult) and were often quite limited in scope; for example, not using a mosquito net the night prior to a survey. This measure, whilst convenient and clearly defined, does not allow a distinction to be made between individuals who never or rarely use a mosquito net, individuals who inconsistently use a mosquito net and individuals who usually use a mosquito net, but for whatever reason did not do so the night prior to survey. Reasons for not using a mosquito net, as well as interventions to encourage greater use, are likely to vary between these three categories of 'non-user'. Thus, the inability to identify the membership of these respective groupings, and their respective reasons for non-use, confounds informed and targeted intervention.
A further limitation of the reviewed reason for mosquito net non-use data was the paucity of qualitative investigation. The qualitative data included in the review were, as with the survey data, typically a minor component of the results presented in the respective studies. The one study in which the reported data were a primary research focus  employed a mixed methodology in which the qualitative component was highly structured, relatively minimal in scope and the level of analysis was limited and poorly described. This study was also distinct in that it examined the reasons why an identified mosquito net went unused as opposed to the reasons why an identified individual did not sleep under a mosquito net. Focusing on the net depersonalizes the line of questioning potentially resulting in honest more accurate data, gives a better sense of which mosquito net types (or states) may be more or less appealing and may allow use of an individual net to be tracked over time. However, by focusing on the net the reasons why the very individuals to whom the mosquito nets are provided do not always use them may go unreported or may be incorrectly reported (if someone other than the individual who would normally sleep under the net responds to the research questioning). The potential loss of, or inaccurate reporting of, relevant data inherent in study designs that focus on the net rather than the individual is of concern as it is these individuals that mosquito net promoting interventions or campaigns must target rather than the nets they choose not to use.
Taken together then, the omission of important information in many of the published survey findings, the constraints on generalization and the dearth of dedicated quantitative and especially qualitative investigation seriously undermine confidence in the reported findings of this review. The seemingly clear patterns evident in the reviewed data and the recommended interventions should, therefore, be considered highly tentative until such time as a greater quantity of dedicated, well designed and reported studies are available in the published literature. The current evidence-base is not sufficient in scope or quality to reliably inform mosquito net promoting interventions or campaigns targeted at individuals who own but do not (reliably) use mosquito nets.
A balanced consideration of the results and recommendations presented above also requires overt acknowledgement of the limitations in the review itself. Grey (unpublished) literature was excluded from review as were the numerous and varied studies that have examined barriers to mosquito net use from perspectives other than that of identified owner/non-users. The latter studies, in particular, often provide instructive data on barriers to mosquito net use and warrant review in their own right. It was the opinion of the authors, however, that the limited scope of this review was justified on the grounds that the published data broadly pertaining to barriers to mosquito net use is so extensive and varied that it is better suited to multiple 'subject specific' reviews rather than a single general review. Other 'barrier to mosquito net' review topics may include: factors predictive of mosquito net use in households with and without sufficient mosquito nets; factors predictive of mosquito net ownership - and the number of mosquito nets owned - following a mass distribution campaign; and focussed investigations into locally specific barriers to mosquito net use.