Ethics principles in the framework

The ethics of voluntary and imposed movement restrictions have been previously discussed for infectious disease outbreaks and other biohazards. In 2001, Barbera and colleagues18 examined the ethical considerations for large-scale quarantine for bioterrorism threats. They primarily focused on assessing the need, feasibility and cost–benefit ratio. Arguably, these ethical considerations for large-scale quarantines18 could be adapted for lockdowns in the ongoing COVID-19 pandemic. In the context of the 2003 severe acute respiratory syndrome epidemic, Upshur discussed ethical principles of harm, proportionality, reciprocity and transparency (see table 1) as requisite justifications for public health interventions,19 and applied them to the imposed and voluntary quarantines.20 In 2007, the WHO’s report on the influenza pandemic discussed the ethical principles to be upheld by governments in case of movement restrictions.21 Beyond those suggested by Upshur, this report also brought forward social justice, liberty, confidentiality, fair process, efficiency and accountability as requirements for ethical imposition of restrictions. In 2015, the Presidential Commission for the Study of Bioethical Issues listed out ethical considerations for restrictive measures in response to the public health planning for the Ebola epidemic.22 The commission report recognised that an ethical implementation should adhere to considerations of the harm, reciprocity, least infringement, evidence-based action proportionality (as beneficence and non-maleficence), justice and fairness, and ensuring equitable benefits sharing across socioeconomic strata. In 2016, the WHO released generic guidance for ethical management of infectious disease outbreaks.23 Recommendations for restrictions on freedom of movement included the justifiable basis for imposing restrictions (known harm and evidence-based action), least restrictive means, cost consideration (utility efficiency), ensuring humane conditions (reciprocal benefits), addressing financial and social consequences, due process protections (fair process), equitable application (fairness and distributive justice), and communication and transparency.

There is a conceptual convergence among these differently named sets of principles and considerations that are unified into an extended list of unique principles (table 1). It is noteworthy that this extended list demonstrates that some of the past discussions have ascribed different notions to a given principle or invoked different principles to present the same idea. For instance, there is a conceptual overlap between the basis of the necessity of intervention and its justifiability. Or that reciprocity has been used to represent the treatment of people at the hands of the authorities and also to denote the obligation of members of a community towards one another. There are also perceivable relationships among the principles. For instance, the harm principle can be considered a precondition for justifiability as without establishing the harm, imposing a restriction cannot be justified. In other words, the evidence underlying to denote harm also acts to affirm the justifiability of the interventions. In another instance, the least restrictive means principle can be thought of as a complement of the proportionality principle where the former concerns the stringency of the restrictions relative to each other, while the latter can be thought to address if the evolution of restrictions matches the trajectory of an outbreak. Reciprocity between the members of a community can have an impact on the equitable distribution of the burden of restriction. Commitment to accountability on the part of decision-makers can engender transparency and mechanisms of checking relevance (eg, conducting public polls on the acceptability of intervention), and can make decision-makers be aware of the problems to initiate a cascade of accountable remedies. Considerations made towards cost and feasibility of implementing a restriction can determine the utility efficiency and vice-versa, while both these principles are conditional to the judgement of necessity. The interdependencies of the principles have a pragmatic value for the policymakers. It is conceivable that upholding or acting towards satisfying all the principles is improbable under the expectation of an urgent response in uncertain circumstances. For instance, limited and uncertain evidence for the magnitude of harm at the onset of an outbreak can make the measurement of harm challenging and thereby impede the judgement over necessity and justifiability. Additionally, the limited and uncertain epidemiological evidence can complicate defending whether or not a restriction is proportional. Hence, at the onset, against the contrast of inability to justify harm or determine the proportionality of the proposed intervention, the ‘ethical focus’ of policymakers should be on complementary principles such as transparency of communication, establishing accountability and conveying the need for reciprocity. Ensuring the basis for least restrictive means and equity becomes important too. As the biomedical, epidemiological and other implementation evidence emerge, policymakers can expand the focus to defending past actions and clarifying future decisions calibrating them according to harm, justifiability and proportionality principles.

Indicators in the framework and considerations for application

Based on the principles that have been unified in the framework (table 1) and the notions represented by them, we first derived actionable considerations or answerable questions. Next, the considerations were mapped onto measurable indicators useful for initiating implementation, monitoring and evaluating restrictions. The importance of principles and the choice of indicators are contingent on the context. The relevance of the policy indicators to a country or local region is determined by, among other factors, the sociocultural acceptability of the corresponding ethical principle. For instance, the demand for the level of transparency along with what counts as transparent communication varies across societies. Hence, although transparency is desired across societies, the indicators for its suitable measurement might at times differ. Most of these indicators are based on easy-to-collect data that are available to the decision-makers. Some specific analytical indicators might require the expertise of the technical (eg, economic) advisors. However, such aid is often available to high-level policymakers and health planners and can also be provided from global collaborations in the absence of local experts.

The current presentation of the framework consolidates the previous discussions around the ethics of movement restrictions as public health interventions (PHIs) into 11 unique principles. The ethical principles are mapped onto 34 policy indicators. For simplicity, 31 out of the 34 indicators (table 1) have been constructed to have dichotomous (yes or present/no or absent) responses of which 3 indicators (analysis of trade-off, matching stringency of measures with the growth of cases and deaths, and sufficient time intervals given for every restrictive step to show its desired effect) are specific to the infectious disease epidemic or pandemic. For instance, in the case of COVID-19, the least restrictive measure was recommended at least a week before the first case.24 However, this recommendation could vary for other pandemics. With evolving knowledge, the proposed framework could incorporate thresholds to dichotomise these indicators. The remaining three indicators (number of steps between the least and most restrictive measures, cost and population coverage of relief measures, and frequency of press conferences) have numerical responses with no normative thresholds. However, higher values for each of these indicators would depict greater adherence to the corresponding ethical principles. For instance, a greater number of steps between the least and most restrictive measures indicates that there was a gradual increase in the restrictions, which corresponds to greater adherence to the ‘least restrictive means’ principle. We do not suggest a fixed scoring system/rules for using the framework. The framework can be applied flexibly with suitable scoring at the discretion of the stakeholders. Generally, anyone with adequate knowledge and the necessary data can use the framework to critique the ethics of movement restrictions. More specifically, it is meant to be a systematic guide for policymakers (eg, local and national pandemic task force members) to ensure ethically sound movement-restricting policies. Before implementation of the movement restrictions, it can be used by the health planners/policymakers/task force for systematically considering the ethics of the movement restrictions. After implementation, it can be used for monitoring whether the ethical aspects of the restrictions are being upheld. It can also be used by independent researchers and policy analysts to assess or critique the ethical soundness of movement restrictions.

Here, we apply the framework to India’s initial national lockdown during the COVID-19 pandemic. In our scoring, we consider that the ethical principles were satisfactorily adhered to if all the dichotomous indicators have an affirmative response (ie, yes) and partially adhered to if more than half the indicators corresponding to the principle had an affirmative response. We did not include the indicators that could not be made dichotomous in this scoring (table 2).

Application of the framework to the national-level Indian lockdown

On 18 January 2020, much before the first COVID-19 case in India, India began airport screening of travellers from China.25 On 11 March, the Disaster Management Act was invoked, all visas were suspended and compulsory quarantine was initiated for all international travellers.25 26 A voluntary curfew was proposed by the prime minister on 22 March and widely popularised by the media.27 India initiated the largest lockdown (here referred to as lockdown 1.0) in the world on 25 March 2020 for 21 days, containing about 1.3 billion people (see online supplemental file 1). Subsequently, it was extended into lockdown 2.0 (15 April–3 May), lockdown 3.0 (4 May–17 May) and lockdown 4.0 (18 May–31 May), followed by phased reopening or unlock periods.28 For a while, some restrictions stood in place, with a push towards safely restarting the economy. However, in light of the disastrous second wave of COVID-19, lockdowns have re-emerged across various Indian states.29 Here, as an example, we apply the framework to lockdowns 1.0 and 2.0 that were imposed by the Central Government of India. We do not include lockdown phases 3.0 and 4.0 as the decision-making became more locally driven during these with individual state governments and authorities playing a greater role, thereby making assessment challenging.

Our findings for Indian lockdown phases 1.0 and 2.0 are summarised in table 2 and extensively described with references to evidence in the online supplemental file 2. We find that most indicators were readily available in the preprint and academic literature and policy documents. The indicators: considering sufficient time intervals for every restrictive step to show maximum effect and the frequency of press conferences were based on simple calculations of extracted data. Of the 31 dichotomous policy indicators, 23 had an affirmative response. Principles of harm, justifiability, transparency, least restrictive means, and cost and feasibility were satisfactorily adhered to, while those of reciprocity and equity were partially adhered to. Proportionality, utility efficiency and accountability were not adhered to according to our scoring system (table 2). It is critical to note that the adherence to ethical principles is contingent on the scoring criteria used here and can be determined differently by other evaluators depending on the context.

Limitations of the framework and its current application

Limitations of the current analysis can be grouped into those relating to the framework itself and those in its current application. With regard to the framework, the current study lacks a validity assessment of the framework. Here, our aim is to introduce the framework and present an example of its application to India’s lockdown. A formal assessment of content and criterion validity needs to be conducted. Even so, to ensure content validity, we mapped several policy indicators to a given ethical principle with the aim to cover multiple facets covered across literature under that principle. Given that there is no ‘gold standard’ for comparison, assessment of criterion validity might be a challenge.

The presence of possible ceiling and flooring effects is an important limitation in the current application of the framework. These effects arise from the use of dichotomised indicators. Categorical indicators, such as dichotomised indicators, are used as they facilitate easy application, interpretation and comparison. In the future, discrete responses or ordinal responses with levels could replace the dichotomisation to avoid potential flooring and ceiling. Further, we used the arbitrary threshold of ‘at least half of the indicators being affirmative’ to decide whether the ethical principle was adhered to. In our analysis, although 23 out of 31 dichotomous indicators had ‘yes’ as an answer, the relative proportion of the indicators within the principles leads to the conclusion that several ethical principles were not completely or partially adhered to. Changing this threshold can lead to qualitatively different conclusions. Here, our attempt was to demonstrate a simple application of the framework hence we did not test the conclusions across different thresholds. Future studies could choose different thresholds deemed suitable for the scenario at hand or better yet, present a range of conclusions sensitive to the different thresholds. While it might increase analytical complexity, studies could also assign differential weights to the indicators or use different methods of aggregation better suited for the assessment at hand. Finally, we applied the framework to the national-level lockdown in India, which simplifies the subnational socioepidemiological heterogeneities and does not consider the variations in the state or further local-level response measures. However, with more granular data, investigators and policymakers at any level in the administrative hierarchy can apply the framework to the administrative geographical unit (eg, state or district or municipality containment zone) of their interest. Such an application will need to use the policy indicators at the multiple decision-making levels, that is, state-level analyses would consider both national and state policies on movement restrictions.