Ethical issues have to be addressed in each step of implementing and running a public health surveillance system. A prerequisite for dealing with ethical issues in an adequate manner is awareness of the full spectrum and complexity of issues that will likely arise. This review therefore provides the first systematic qualitative review of the literature for those developing guidance and training material, as well as public health practitioners. As our target audience were practitioners and not (foremost) the bioethics community, we found it most helpful to use the phases of surveillance for synthesising the data and not, for example, normative principles. We could thereby ground the analysis in the practical reality of those working in the field. Our review is furthermore different from other reviews of ethical issues [18, 28] in that it also extracted proposed strategies to adequately deal with ethical issues identified, although we only present our findings on the issue of informed consent. From the standpoint of guideline developers, this broader perspective seems desirable as it allows formulating recommendations regarding how to handle diverse issues knowing all arguments put forward in the literature. This is the first study to demonstrate the feasibility of such an approach. In the future, to maximize the instrumental value for practitioners, guideline developers, and other users of analyses of this kind, systematic (qualitative) review approaches should be further refined based on feedback from users.
Generally, guidelines for conducting systematic reviews require researchers to assess the quality of the data included . However, the notion of ‘quality’ is underdeveloped for ethics inquiries and we therefore refrained from any quality assessment. This also means we did not assess the practical relevance of ethical issues – one promising candidate for operationalizing quality in this context. The number of articles devoted to one issue does not function as a reliable indicator of importance due a number of potential limiting factors (e.g. publication bias), which was also the reason why we did not count the number of times a particular code was discussed in the literature. Moreover, we did not address the normative relevance of justificatory conditions introduced in the literature. Decision-makers will have to weigh the different arguments within and for their specific contexts.
Nevertheless, we want to raise a few comments with regards to the quality and breadth of the scientific discourse in the field of public health surveillance. First, we found it critical that most theoretical articles we read did not substantiate real-life risks described with empirical data that could have demonstrated their relevance. To give but one example: stigmatization as a result of surveillance is of course a theoretical possibility, but it is hardly discussed how often, in what setting, or for which diseases stigmatization and discrimination are actually experienced as a consequence of published surveillance data. Judging from the literature we have read, the relevant empirical literature is not adequately taken into account in normative documents though authors’ normative arguments often rested on empirical premises. Normative scholars in public health should therefore make a point of considering the relevant empirical literature more explicitly.
It was furthermore noteworthy that among the publications included only one focused specifically on a low- income country  and two focused on a method of data collection (verbal autopsy) that is generally only employed in a low-income context [36, 37]. Furthermore, five papers focused on global public health surveillance, which actually included the low- and middle-income countries’ perspective [38–42]. However, the majority of articles focused on privacy in digitalized surveillance systems, which may be of greater importance in high-income countries. It seems that the perspective of low-income countries is not well represented in the literature possibly due to preferences of researchers, funding opportunities or publication bias. However, we do not think that this is particular for the field of surveillance research, but rather represents a general imbalance in knowledge generation that tends to focus on financially strong contexts. As to be expected, there are no ready-made solutions for this problem.
Regarding the discussion of justificatory conditions, we found it questionable that many articles stayed on a very abstract level. Often authors just restated the – rather abstract and broad – justificatory conditions already outlined in the framework of Childress et al. . It is not enough to just say that the benefits of surveillance need to be considerable to justify foregoing informed consent, practitioners need to know what would constitute a (in-) considerable benefit. It would be desirable if more authors could like Lee et al.  at least provide examples (in her case HIV where limited data could lead to underfunding of programs, increase infections and ultimately death or so she claims) and rationales for why certain justificatory conditions are fulfilled in a particular context. Ethicists motivated to improve public health practice should not shy away from more context-specific analyses, thereby also providing example cases of normative deliberations that practitioners will inevitably have to address.
Lastly, we want to emphasize again that systematic reviews can only be a starting point for any policy process. Other types of ethical inquiry – like commissioned evaluation of arguments or analysis of ethical issues in novel areas – might and possibly should supplement our research as information base for decision-making. Given that the literature will always be limited due to personal preferences, funding priorities and scientific paradigms, it has been recommended [20, 43] that systematic review findings are accompanied by hearings or surveys of experts and possibly affected communities. This will ensure that all important issues and arguments are captured and discussed during guideline or further policy development.
One limitation of this review is the inclusion of only two databases (PubMed and Google Books). However, those can be considered the most important ones, allowing us to include both books and journal articles. Previous systematic reviews of medical ethics literature have shown only marginal additional value of searching other databases like EMBASE or Euroethics [19, 44]. We could have also included additional terms to our search strategy, for example, data sources for surveillance like “screening” or “health surveys”. However, this would have increased the number of hits for us to screen considerably while promising only few additional relevant findings. Furthermore we did not have to adapt our matrix on the level of themes and only had to make marginal changes on the level of codes for the last round of analysis of the final 35 publications. We therefore assumed thematic or conceptual saturation for the level of codes, meaning that analyzing further documents would not reveal other ethical issues at the abstraction level of codes . Due to language capacities within the research team we could only include articles published in English, German, and Spanish. However, only two papers (one written in French  and one in Portuguese ) were excluded based on language. We are therefore confident that we have captured the majority of issues.
One could criticize how we operationalized the fundamental but contested concepts that grounded our search, first and foremost how we defined what constitutes an ethical issue. First, we based our approach on principlism. This lens might have complicated capturing more postmodern or constructivist discussions and some issues that can arguably be described as ethical will not have been identified here. One example would be discussions that try to clarify a particular concept – e.g. privacy rights and their normative justification – or discussions that do not refer to e.g. harm or autonomy violations, but more diffuse notions of the good to criticize certain actions. However, we had good reasons for choosing an approach based on principlism. Most importantly, other systematic qualitative reviews of normative information already demonstrated the instrumental value of this approach for a descriptive and stakeholder-oriented analysis and synthesis of normative arguments [18, 19, 28]. Additionally, we found this approach to be more grounded in the actual theoretical discussions in the public health ethics literature than, for example, an approach relying on the authors’ statement in the abstract to discuss ethical issues (which other reviews relied on ). Second, we also chose one particular principlist framework over possible alternatives . It is difficult to say how choice of a different framework might have impacted our analysis, but we are convinced that the inclusive nature of the chosen framework with regards to principles translated to inclusiveness of our research with regards to issues and arguments. Further research still has to demonstrate whether alternative approaches result in equal, superior, or inferior results.
Additionally, systematic qualitative ethics reviews involve a high level of interpretation because authors do not always clearly describe the issues at hand, thereby introducing a subjective or interpretive element to the analysis on the part of those conducting the review. However, three authors were involved in reading and analysing the literature and met regularly to discuss challenges with interpretations to ensure the reliability and validity of the findings. We are therefore confident that our interpretations properly represent the data.