UK case law

Janis Sostaks & Anor v Prosecutor General’s Office, Latvia

[2025] EWHC ADMIN 2795 · High Court (Administrative Court) · 2025

Get your free legal insight →Email to a colleague
Get your free legal insight on this case →

The verbatim text of this UK judgment. Sourced directly from The National Archives Find Case Law. Not an AI summary, not a paraphrase — every word below is the original ruling, under Crown copyright and the Open Government Licence v3.0.

Full judgment

Introduction

1. This judgment addresses, not for the first time, the question of prison conditions in Latvia, a Council of Europe state to which the United Kingdom extradites criminals convicted there or accused of crimes and wanted to stand trial there. The arrangements in this country are found in Part 1 of the Extradition Act 2003 ( ). Latvia is a European Union ( the 2003 Act EU ) state and a category 1 territory under that Act . Internationally, the arrangements are found in Part 3, Title VII of the Trade and Cooperation Agreement of 30 December 2020 ( TACA ). In full, the Trade and Cooperation Agreement between the European Union and the European Atomic Energy Community, of the one part, and the United Kingdom of Great Britain and Northern Ireland, of the other part.

2. The appellant, Mr Sostaks, and the applicant, Mr Igoniņs, are in custody here and currently receive opioid substitution therapy at HMP Wandsworth for their long term substance abuse and addiction. Both say there is a real risk that their detention in the Latvia would lead to them receiving degrading treatment in breach of article 3 of the ECHR and they should therefore be discharged under section 21 A(4)(a) of the 2003 Act . Alternatively, they ask the court to seek appropriate assurances from the Latvian authorities under TACA article 604(c).

3. Mr Sostaks has permission from Sweeting J to appeal against the order of the Deputy Chief Magistrate, District Judge Ikram, made on 10 October 2024, to extradite him to Latvia to face trial on five charges of theft, alleged in an accusation warrant to have been committed in Daugavpils from February to June 2021. He has Fordham J’s permission to advance a further ground: that his extradition would be disproportionate under section 21 A(1)(b) of the 2003 Act , principally because of the deductible time served in HMP Wandsworth.

4. Mr Igoniņs’ extradition to Latvia was ordered by District Judge Goozée on 29 November 2024 for reasons given in the judgment of District Judge Clarke in her judgment the same day. He is wanted to stand trial for two drugs offences alleged in an accusation warrant: buying and storing a “large quantity” of narcotic substance, in Riga in about April 2019; and buying and storing a psychotropic substance, carfentanyl, in Riga in about June 2019. McGowan J ordered a rolled up hearing in his case and that it should be joined to Mr Sostaks’ appeal.

5. I had full written and oral argument from counsel for Messrs Sostaks and Igoniņs, and for the respondent. I grant Mr Igoniņs permission to appeal on the article 3 ground. Both appellants also sought to advance a challenge to the decisions below based on article 8 of the ECHR, but both have sensibly abandoned that ground of appeal in the light of the Supreme Court’s decision in Andrysiewicz v. Circuit Court in Lodz, Poland [2025] 1 WLR 2733 .

6. The appellants say they will suffer degrading treatment if they attempt to obtain treatment for their drug addiction within the Latvian prison regime: those who do so become so-called “untouchables” within an informal prisoner hierarchy which divides prisoners into castes. The appellants must, they submit, either forego drug rehabilitation treatment, normally a pre-condition of early release, or suffer the degrading treatment meted out to untouchables in violation of ECHR article 3. The Facts

7. My chronological account includes case law documenting the state of Latvian prisons. Mr Igoniņs was born in July 1992 and is now 33. Mr Sostaks was born in January 1995 and is now 30. Both are Latvian nationals. Both have drug addiction problems. Both have convictions in Latvia and in this country for offending associated with drug addiction. Both came here while wanted by the Latvian authorities for the offences alleged in the accusation warrants. Further details need not be set out here, though I shall return to Mr Sostaks’ record later when considering his second ground of appeal.

8. The issue of prison conditions in Latvia has been considered in several cases including DF v. Latvia (app. no. 11160/07). The European Court of Human Rights decided in October 2013 that DF’s article 3 rights had been violated during his imprisonment which started in 2005. He claimed to be a police informer and had suffered violence from other prisoners. It was common ground that sex offenders and police informers were at heightened risk of violence from other prisoners. His transfer to a different prison was delayed, a matter in respect of which he had no effective domestic remedy, the court found.

9. When the Strasbourg court decided DF , article 3 arguments were also being raised in this country in numerous cases concerning Latvian prison conditions. Three appeals listed together were heard by Moses LJ and Collins J in March 2014. In their judgment in April 2014 they rejected any article 3 challenge, dismissed all three appeals and made clear that seven similar challenges then pending before the magistrates’ court were doomed to fail: see Brazuks v. Prosecutor General’s Office, Republic of Latvia [2014] EWHC 1021 (Admin) .

10. The outcome was the same in Danfelds v. General Prosecutor’s Office, Latvia [2020] EWHC 3199 (Admin) . Popplewell LJ and Johnson J heard two appeals together. Certain assurances had been sought at permission stage. Information in response was provided in August 2020. At the substantive stage, the court considered that material and other evidence including a 2019 report from the Committee on Prevention of Torture ( CPT ) but upheld the decisions below, adding at [65] that evidence “falls far short of what would be required to rebut the presumption that Latvia complies with its obligations under Article 3 …. .”

11. A delegation of the CPT visited three Latvian prisons (among other institutions) in May 2022 and published its resulting report over a year later, on 11 July 2023 ( the 2023 CPT report ). There were positive comments but the summary said the CPT was “seriously concerned to note that no significant progress had been made in reducing the scale of Or more simply, “the report”. It is sometimes referred to as the “2022” CPT report. I have opted for the year of publication not the year the visit took place. inter-prisoner violence ” (underlining in original); and that “[t]he informal prisoner hierarchy (or caste system) still seemed to be key foundation of prisoners’ life in the three prisons visited, with its traditions dictating internal order and being given priority over official rules.”

12. In more detail, paragraphs 71-81 on inter-prisoner violence stated that the committee had heard many credible allegations of such violence, including beatings and psychological pressure, though the level had decreased at Riga Central Prison. The problem remained serious at Jelgava and Daugavrīva Prisons. The cause was a combination of informal prisoner hierarchies, insufficient staff presence in accommodation areas and idleness. Some of the incidents were reported but victims would claim to have sustained the injuries accidentally, for fear of reprisals; this made the statistics unreliable.

13. The report went on to explain that some inmates would refuse a transfer to more open conditions controlled by those at the top of the informal hierarchy. Those in the “lowest caste” – the “untouchables” - were subjected to threats, verbal abuse and demeaning behaviour, being compelled to perform dirty work such as cleaning toilets and collecting rubbish. Informal leaders enjoyed higher quality furniture and living conditions. The untouchables had to obey orders from higher caste prisoners and not mix with them. Prison staff did not deny the problem and assured the delegation they were trying to tackle it.

14. The committee’s view was that “the situation of ‘untouchable’ prisoners in Latvia could be considered to be a continuing violation of Article 3 …. .” (paragraph 77). The CPT called on the Latvian authorities to take resolute action without further delay to address the “systemic and present shortcomings” in the prison establishment including “pro-active steps to combat inter-prisoner violence” by investing resources, recruiting more staff and offering decent living conditions (paragraph 80).

15. At paragraph 81 the CPT considered drug rehabilitation (emphases in original): “Another area of concern, related to the phenomenon of informal prisoner hierarchy, was the reluctance of prisoners, in all establishments visited but more specifically at Daugavgrīva Prison, to participate in the rehabilitation programme in Olaine Prison Hospital’s Centre for inmates having had experience with addiction . The prisons’ managements confirmed that going through such a programme was almost the only channel through which a change of sentencing regime could become possible. In a later stage, it also influenced potential access to early release. [ footnote 62 ] As the management of Daugavgrīva Prison was also well aware, inmates coming back from Olaine Rehabilitation Centre had to face threats to their physical and psychological integrity because they were systematically considered as being part of the lowest caste of the hierarchy. This was the reason why very few prisoners accepted to go through the rehabilitation programme at Olaine (and therefore remained on the low-level regime). The CPT would like to receive the data on the number of prisoners who have completed the rehabilitation programme at Olaine Rehabilitation Centre since January 2019, as well as the proportion of prisoners who were transferred from the low- to high-level regime in the same period of time. The CPT welcomes the project of a “drug-free zone”, which was recently launched and implemented in Block 4 at Grīva Section, allowing prisoners coming back from such a rehabilitation programme not to have to be mixed with the general prison population; it therefore ensured them to have access to a safe place upon return. This type of initiative should also be encouraged in other prison establishments. ”

16. At footnote 62 the report stated: “[j]udges deciding on early release made it conditional upon the prisoner having completed the full rehabilitation programme at Olaine Rehabilitation Centre (based on Section 61 of the Criminal Code).”

17. The Latvian government’s response to the 2023 CPT report was published the same day. It acknowledged the problem of inter-prisoner violence and said that work was underway to introduce specific “resocialisation programmes developing social, communication and interaction skills, discovering and correcting thinking errors.” Three particular “Reasoning and Rehabilitation” programmes were at the planning stage and were to be introduced (page 26). The Latvian Prison Administration ( LPA ) “undertakes to take steps to address the shortcomings outlined in the Report and previous reports … .”

18. In response to paragraph 81 of the 2023 CPT report, the government stated that from January 2019 to December 2022, the addiction reduction programme at the Addiction Centre (at Olaine) had been completed by 270 convicted persons. The sentencing regime had been “alleviated” during the same period in 1,551 cases. Undergoing drug rehabilitation is one of various criteria for “alleviating” the sentencing regime, i.e. early release or a reduced term. Participation in other activities showing “resocialisation” is also relevant: employment, education, psychological aid, social problem solving, leisure time activity and other resocialisation measures are also taken into account (page 28).

19. On 14 November 2023, Swift J gave judgment in Vascenkovs v Prosecutor General’s Office, Republic of Latvia [2023] EWHC 2830 (Admin) . The appeal concerned an accusation warrant relating to alleged benefit fraud. At the substantive hearing of an appeal on two grounds, the appellant sought permission to add ECHR article 3 as an additional third ground. Swift J refused to allow the amendment for reasons he gave at [32]-[36]. He referred to Danfelds , cited passages from the 2023 CPT report on inter-prisoner violence and some passages on the same subject in the Latvian government’s response.

20. Swift J rejected in short order ([35]-[36]) the submission that “matters have moved on” since an earlier 2017 CPT report considered in Danfelds . He referred to the strong presumption of compliance which “will prevail save where exceptional circumstances are demonstrated”. He referred to the Latvian government’s response and to “the specific steps being taken to re-socialise prisoners” and “to increase the number of prison officers”. While the slow progress since 2016 was of concern, the situation had not deteriorated since then and the authorities remained committed to addressing the shortcomings identified by the CPT and appeared to have made some progress at Riga prison.

21. Three days later, as it happened, Mr Igoniņs was arrested in Cambridgeshire on the basis of the accusation warrant in his case. He has been in custody at HMP Wandsworth, receiving opioid substitution therapy, since then. At the same time, the Court of Human Rights in Strasbourg was considering a further case about prison conditions in Latvia, D. v. Latvia (app. no. 76680/17) filed in 2017 by a Latvian prisoner who had served a sentence in three Latvian prisons from 2008 to 2017. The complainant, D, had exhausted national remedies in May 2017 without success and then filed his complaint to the Strasbourg court in October 2017.

22. The judgment in that case was issued on 11 January 2024, as it happened a month before Mr Sostaks was arrested in Watford based on the warrant in his case, whereupon he joined Mr Igoniņs at the same prison, with the same therapy. In D. v. Latvia , the Strasbourg court elaborated on the caste system, explaining at [6] that there were three castes: “the “ blatnie ” (the highest caste), the “ mužiki ” (the middle caste), and the “ kreisie ” (the lowest caste).” The applicant, D, was of the kreisie , being a sex offender. The kreisie , the court noted at [7]: “were not permitted to sit on the same benches, nor to walk or stand in the same areas as other prisoners. Specific toilets, sinks, and dining areas were designated solely for their use. Moreover, they were prohibited from mingling with other inmates in the queue to the prison shop or the medical department. Participation in communal sporting activities or sharing shower facilities was also forbidden. They had to sleep on the least comfortable beds, located towards the edges of the living areas, and could not go into areas reserved for the other two castes. In addition to these restrictions, inmates of the lowest caste were tasked with performing undesirable or menial labour on behalf of other prisoners, which included standing guard, cleaning rooms, and laundering clothes.”

23. The European Court of Human Rights had before it reports from the Latvian Ombudsman and reports of the CPT up to and including the 2023 CPT report (from which at [29] the court cited paragraph 77, also mentioned above). As this is (so far as I am aware) the first domestic appeal about Latvian prison conditions since D. v. Latvia , it is worth quoting from the judgment in extenso . It is a good account of the law on article 3 in this context: ‘ (c) Whether the applicant was subjected to a treatment prohibited by Article 3

46. The Court reiterates that ill-treatment must attain a minimum level of severity if it is to fall within the scope of Article 3 of the Convention. Ill-treatment that attains such a minimum level of severity usually involves actual bodily injury or intense physical or mental suffering. However, Article 3 cannot be limited to acts of physical ill-treatment; it also covers the infliction of psychological suffering. Where treatment humiliates or debases an individual, showing a lack of respect for or diminishing his or her human dignity, or arouses feelings of fear, anguish or inferiority capable of breaking an individual’s moral and physical resistance, it may be characterised as degrading and also fall within the prohibition of Article 3 (see Ananyev and Others , cited above, § 140, and Begheluri v. Georgia , no. 28490/02, § 100, 7 October 2014). The Court further reiterates that acts of abuse other than physical violence may also constitute ill-treatment because of the psychological harm they cause to human dignity. In particular, a threat of ill-treatment can also amount to a form of ill-treatment because of the fear of violence it instils in the victim and the mental suffering it entails (see Gäfgen v. Germany [GC], no. 22978/05, § 108, ECHR 2010).

47. With regard to what constitutes that minimum level of severity, the Court’s approach has evolved. Initially, the Court held that “the mere feeling of stress of a detained person” (see I.T. v. Romania (dec.), no. 40155/02, 24 November 2005) or “the mere fear of reprisals from the [applicant’s] cellmates” (see Golubev v. Russia (dec.), no. 26260/02, 9 November 2006) were not of themselves sufficient to bring the situation within the ambit of Article 3. However, when this fear of reprisals was accompanied by other factors, the Court has found that “the cumulative effect of overcrowding and the intentional placement of a person in a cell with persons who may present a danger to him may in principle raise an issue under Article 3 of the Convention” (see Gorea v. Moldova , no. 21984/05, § 47, 17 July 2007). In two subsequent cases, Rodić and Others v. Bosnia and Herzegovina (no. 22893/05, § 73, 27 May 2008), and Alexandru Marius Radu v. Romania (no. 34022/05, § 48, 21 July 2009), the Court found a violation of Article 3 on the grounds that “the hardship the applicants endured, in particular the constant mental anxiety caused by the threat of physical violence and the anticipation of such ... must have exceeded the unavoidable level [of suffering] inherent in detention”. Finally, in the cases of vulnerable prisoners in Latvian prisons, the Court held that a year-long exposure to “the protracted fear and anguish of the imminent risk of ill-treatment”, coupled with the absence of an effective domestic remedy, amounted to a violation of Article 3 (see D.F. v. Latvia , cited above, § 95, and compare J.L. v. Latvia , cited above, §§ 74-75 and 87-88).

48. In a recent case addressing the issue of an informal hierarchy among prisoners, the Court made significant findings regarding the threshold of severity triggering the application of Article 3 of the Convention. Although not all applicants in that case categorised as “outcast” prisoners experienced physical violence, they nonetheless lived under a constant threat of such violence for breaching the informal regulations. The resulting mental anguish and fear of ill-treatment were considered to undermine their human dignity and instil a sense of inferiority in them, thereby constituting a form of degrading treatment in violation of Article 3 (see S.P. and Others , cited above, § 92). The Court elaborated that the restrictions endured by the “outcast” applicants served as additional evidence of degrading treatment. Their separation from other inmates occurred on both physical and symbolic levels; they were allocated less comfortable spaces in the dormitory and canteen and had restricted access to essential prison resources like showers and medical care. Furthermore, they were not allowed to come into proximity with, much less touch, other inmates. This denial of human contact led to their social isolation and likely caused significant psychological repercussions (ibid., § 93). In addition, “outcast” applicants were forced to perform work considered to be inherently degrading and unacceptable for the other prisoners. This further debased them and perpetuated the feelings of inferiority (ibid., § 94).

49. In the present case, the applicant’s account bears relevant similarity to the above case, specifically as regards the physical and symbolic separation faced by prisoners in the " kreisie " category to which he belonged. He indicated that prisoners in that lowest group faced many arbitrary restrictions on using shared resources. They had separate benches, toilets, and dining areas and were not allowed to queue with other prisoners for the shop or medical care. They were also banned from joining in sports or using common showers. Their beds were less comfortable and located towards the periphery of shared spaces. In addition, they were tasked with performing menial jobs, such as cleaning and doing laundry for the other inmates (see paragraph 7 above). The Court concludes that such physical and symbolic separation has had the effect of sending a potent message of inferiority, thereby undermining the human dignity of prisoners in the applicant’s situation, and thus constitutes degrading treatment within the meaning of Article 3.

50. The fact that the applicant chose to comply with the demands and limitations set by the informal hierarchy, rather than opposing or challenging them, does not undermine the Court’s conclusion regarding the degrading nature of these dehumanising practices. The applicant’s emphasis on his own resilience, rather than on the tangible effects of the hierarchical norms imposed, provides the Court with insight into the coping mechanisms that prisoners in his situation may employ. While such mechanisms could potentially mask the full extent of emotional distress, it is imperative to recognise that the lack of overt confrontation and violent incidents does not lessen the reality of the underlying suffering. Life in such a hostile environment often results in a continuous accumulation of stress, particularly for individuals subjected to inequity, and not solely from immediate or chronic threats. The mere anticipation of such threats can also cause enduring mental harm and anxiety of an intensity exceeding the level of stress caused by detention under normal conditions.

51. In the light of the above, the Court finds that the applicant’s physical and social segregation, coupled with restricted access to basic prison resources and denial of human contact, has led him to endure mental anxiety that must have exceeded the unavoidable level of suffering inherent in detention, even though he has not been subjected to physical violence (compare S.P. and Others , cited above, § 96). That situation which he endured for years on account of his position in the lowest caste of prisoners in an informal hierarchy amounted to a treatment prohibited under Article 3 of the Convention. Since the applicant was personally affected by that situation, his claim cannot be said to amount to actio popularis .

52. It remains to be determined whether the domestic authorities have adequately addressed the problem. (d) State’s obligation to protect the applicant from ill-treatment

53. The Court established above that the applicant did not experience any ill-treatment from prison staff. Nevertheless, the absence of any direct State involvement in acts of ill-treatment that meet the condition of severity such as to engage Article 3 of the Convention does not absolve the State from its obligations under this provision (see Gjini v. Serbia , no. 1128/16, § 77, 15 January 2019). In this connection, the Court refers to the relevant principles concerning State responsibility, supervision and control in relation to detention, as well as the obligation to protect an individual from inter‑prisoner violence, which are set out in the case of Premininy (cited above, §§ 82-88). In particular, the national authorities have an obligation to take measures to ensure that individuals within their jurisdiction are not subjected to torture or to inhuman or degrading treatment or punishment, including such ill-treatment administered by private individuals (ibid., § 84, and D.F. v. Latvia , cited above, §§ 83-84). The extent of this obligation of protection depends on the particular circumstances of each case (see Stasi , cited above, § 79).

54. Turning to the present case, the Court observes that the issue of informal prisoner hierarchies is well-documented within Latvian penal institutions. The applicant indicated, and the Government did not dispute, that this problem has been prevalent in the prisons where he was held (see paragraphs 5-7 above). In this situation, both the prison staff and the broader authorities should reasonably have been aware not only of the existence of the hierarchy but also of the applicant’s subordinate position within it. Even in the absence of explicit incidents of violence or confrontation, the inherent risk of ill-treatment faced by the applicant during his term of imprisonment could not be overlooked. Since the authorities were apparently aware of the risk confronting the applicant in this vulnerable situation, it falls to the Government to explain the measures the domestic authorities have implemented to mitigate the applicant’s vulnerability and to address the broader issue of prisoner hierarchies (see D.F. v. Latvia , cited above, § 87, and S.P. and Others , cited above, § 99).

55. The Court reiterates that the complaints concerning the degrading effects of an informal prisoner hierarchy are similar to other complaints that arise from structural problems in a prison environment. Such problems indicate a systemic failure rather than issues isolated to the specific circumstances of an individual applicant (see S.P. and Others , cited above, § 103). Given the systemic nature of these issues, individual interventions – such as initiating an inquiry or transferring the applicant to a different cell or facility – would not have addressed the core issue at the heart of the applicant’s grievances. Even if specific incidents of violence or ill-treatment were to be investigated and those responsible held to account, such measures would not alter the underlying power dynamics of the informal prisoner hierarchy, nor would they change the applicant’s subordinate position within it (ibid., § 104).

56. Limiting interventions to addressing specific incidents, as and when they arise, does not constitute the comprehensive approach that prison management authorities should adopt when grappling with a systemic issue such as an informal prisoner hierarchy. Since 2012 the Latvian Ombudsman has consistently criticised the lack of such an overarching approach, underlining the shortcomings of the traditional, incident-focused strategy (see paragraph 28 above). In the 2018 Report, the Ombudsman went further by asserting that the domestic authorities are acting unlawfully in their failure to dismantle the established informal hierarchies within prisons. This perspective was echoed by the Supreme Court of Latvia, which opined that State authorities bear a general obligation to address the issue of informal hierarchies, which is one contributing factor to inter-prisoner violence (see paragraph 26 above).

57. The Court accordingly finds that the domestic authorities have not taken adequate steps to protect the applicant from the treatment associated with his belonging to the group of “ kreisie ” prisoners. The domestic authorities did not have in place effective mechanisms to improve the applicant’s individual situation or to deal with the issue in a comprehensive manner.

58. There has accordingly been a violation of Article 3 of the Convention on account of the State authorities’ failure to protect the applicant from the treatment prohibited under that provision. Having reached this finding, the Court deems it unnecessary to examine the same set of facts from the perspective of Article 14 of the Convention.’

24. I should add that at the end of its judgment, the court concluded ([62]): “62. The Court considers that to prevent future similar violations, the domestic authorities must address the issue of informal prison hierarchies highlighted in this judgment in a manner that goes beyond the circumstances of the present case. It falls to the competent authorities, in accordance with the respondent State’s obligations under Article 46 of the Convention, to draw the necessary conclusions from the present judgment and to take appropriate general measures in order to address the problem that has led to the finding of a violation here. More specifically, the domestic courts are required to take due account of the Convention standards as applied in this judgment (compare, for a similar approach, Yüksel Yalçınkaya v. Türkiye [GC], no. 15669/20, § 418, 26 September 2023).”

25. A few months after that judgment, from 22 to 31 May 2024 the CPT made a further nine day visit to four prisons in Latvia, Daugavgrīva, Jēkabpils, Jelgava, and Valmiera. The purpose was (as per the subsequent report, not published till February 2025 ( the 2025 CPT report )) “to examine the treatment of persons held in prisons and to follow up on the situation found in the 2022 visit” (page 3). After that visit but before publication of the 2025 CPT report, and before publication also in February 2025 of the Latvian government’s response, the final hearing in Mr Sostaks’ case took place in Westminster Magistrates’ Court. Or, “the report”. It is sometimes referred to as the “2024” CPT report. Again, I prefer the year of publication to the year the visit took place.

26. That hearing was on 23 September 2024. District Judge Ikram gave his succinct reserved judgment on 10 October 2024. He set out the law on article 3 based claims in uncontroversial terms, referring to case law I need not rehearse here. He stated the proposition that the burden is on the requested person to demonstrate that there are strong grounds for believing that if returned, he will face a real risk of being subjected to torture or inhuman or degrading treatment or punishment. The risk must be substantial and not merely fanciful.

27. He noted that it was necessary to examine the foreseeable consequences of sending the person to the receiving state, bearing in mind the general situation and his personal circumstances. The existence of human rights violations in the receiving state was not the same as the risk that a particular individual would be at risk of being subjected to those violations, he noted. That would depend on whether the individual was specifically vulnerable to human rights abuse because of a particular characteristic.

28. The judge then referred to the strong presumption that a Council of Europe member state such as Latvia would be able and willing to fulfil its obligations under the ECHR; and that the presumption was even stronger in the case of an EU state such as Latvia which participates in the European Arrest Warrant scheme. The strength of the presumption is unaffected by the United Kingdom having left the EU.

29. The judge had an expert report before him. In his conclusions on article 3, he said this: “5. The expert report is based on sources of information listed at page 2. The expert states that a likely sentence includes ‘deprivation of liberty.’ The RP has previous offences / convictions in Latvia.

6. The expert states that the RP would not be entitled to a status entitling him to ‘special procedure protection’ because [s]he says that the threats to him arise of out of financial debt rather than his testimony. The JA call no evidence in rebuttal and did not ask the witness be called to give live evidence to be cross examined. I accept what [s]he says. I find that he will not be afforded that special status which appears to be available only on the basis of threats connected with a person giving testimony. The lack of special status does not in itself, undermine the overall presumption that the authorities will provide a prison environment that is Article 3 compliant. The expert states that he will be able to turn to law enforcement agencies as he would be able to here.

7. Prison conditions and his encounters in 2019/2020 are not evidence of prison conditions today and prison regimes including security. The authorities will be well aware of concerns raised on inter prisoners violence (they are not unique to Latvia and are, incidentally, a real issue here in the UK) and can properly be assumed to be addressing the same.

8. I am bound to reject this argument.”

30. The judge also rejected the argument that it would be disproportionate, having regard to the factors in section 21 A(2)(a)-(c) of the 2003 Act , to order Mr Sostaks’ extradition. He concluded as follows: “I am satisfied that whilst the alleged offences are not, individually, the most serious, they form part of a series of offences over a relatively short period. They were also allegedly committed after he had been convicted of other offences of dishonesty in Latvia. Consideration of seriousness and likely penalty / sentence must take into account other relevant offending which must, in this case, have predated at least some of these allegations. They could well attract a custodial sentence in this country.”

31. In Mr Igoniņs’ case, the final hearing before District Judge Clarke was on 1 November 2024. The judge gave judgment on 29 November 2024. I need only refer to parts of it. Mr Igoniņs’ evidence included his claim to be addicted to fentanyl, a police informer and in debt to drug dealers. The judge recorded that he was asked in cross-examination about his evidence that he would become an “untouchable” if he were extradited to Latvia, if he sought drug rehabilitation treatment to gain early release after serving half or two thirds of his sentence.

32. The judge found that he was a fugitive from Latvian justice for reasons it is not necessary to recite. She referred to the 2023 CPT report and the response of the Latvian government. She had before her the relevant case law including D. v. Latvia . At [49] she stated that she could not find “that everyone who is in the lowest caste, and considered an untouchable, is subject to Article 3 non-compliant conditions, however I note that the [2023] CPT report that those in such a situation could [italics in original] be subject to such conditions.”

33. The judge discussed the case law at [50]-[61], uncontroversially. She decided the presumption of compliance by Latvia with its obligations under article 3 was not rebutted. At [62] the judge noted that when serving a previous sentence in Latvia the evidence of Mr Igoniņs was that he had been a mužik , i.e. in the middle caste, not an untouchable. The judge accepted that the 2023 CPT report provided support for the proposition that he would become an untouchable if he took part in the drug programme in prison. She accepted it was “highly likely” he would need to complete that course before successfully applying for early release.

34. She reasoned at [62] that Mr Igoniņs could choose between undertaking the course and securing early release at the cost of becoming an untouchable, or not taking the course, remaining a “middle tier” prisoner (i.e. a mužik ), making early release highly unlikely. The core of her reasoning then followed: “I have not been provided any authority which suggests that serving the entirety of a sentence amounts to Article 3 not compliant treatment. I also do not have any evidence before me which would suggest that the RP cannot be provided, or will not be provided with medical assistance with his drug addiction. His own evidence is that he was able to take part in some work with groups for those with addictions, which was not part of the drug programme. He also gave evidence that there was a nurse in the prison, as one would expect, and he was able to seek the nurses help although he did not seek it specifically for his drug addiction. Whilst in custody in this jurisdiction he is able to get a prescription for methadone. It has not been asserted and nor have I been provided with evidence to support an assertion that he would not be able to get access to such a prescription in the prison system in Latvia without undertaking the drug programme. Therefore, if the RP does not undertake the programme, he will not get early release, he will also not fall into the bottom category off the hierarchy and will not experience the difficulties outlined by the CPT.”

35. After that, on 26 February 2025 the CPT published the 2025 CPT report arising from their May 2024 return visit to Latvian prisons. The main points to emerge were as follows. There was a lack of progress in converting dormitory accommodation to cellular accommodation and with suppressing the culture of the informal prisoner hierarchy. It remained “deeply embedded in almost every aspect of daily prison life.” Complaints of ill-treatment were still rare and often not properly investigated. There were not enough prison staff. Among the lowest caste members (now also called levyje , see paragraph 45 and footnotes) there were some cases that could amount to modern slavery. Their treatment “could be considered to constitute a continuing violation of Article 3… .”

36. The report then went into more detail. The prison population was 3,300 persons, representing 68.2 per cent occupancy. There was therefore no overcrowding. A new prison was being built with cellular accommodation, two to a cell throughout. It was due to open in 2026. The CPT called on the Latvian authorities to speed up progress with conversion to cellular accommodation, suppression of the informal hierarchy and inter-prisoner violence, to increase staff and improve assistance to prisoners with substance abuse problems.

37. Most prisoners interviewed said the rules of the informal hierarchy were “slowly changing, becoming vaguer, and the pressure to respect them becoming less intense”, though breaching them could still lead to physical punishment, often in a way that leaves no visible injury, such as beating the upper body rather than hitting the face (paragraphs 53-54). There were changes: some blatnyje (the CPT’s spelling) were voluntarily demoted to levyje , finding the higher office burdensome or being fed up with the “stupid rules” (paragraph 56).

38. In support of its view that the levyje were receiving treatment in continuing breach of article 3, the report authors cited D. v. Latvia and mentioned that the delegation had been told by the prison authorities during the visit that the action plan to implement measures in response to the judgment would be approved by the government of Latvia in October 2024 (paragraph 60). The CPT expressed the hope that the measures suggested in the 2025 CPT report would be included in that action plan.

39. The report then addressed the subject of illicit drug use. Drugs in the prisons were rife, they said. The drugs arrive in many ways, from drone drops to smuggling in lorries and even by crossbow (paragraph 61). Upper and middle caste members insisted that the sale of drugs between inmates was prohibited; they could only be shared for free. A vor v zakone or “thief in law”, i.e. a leading prisoner, forbad as immoral making profit from another’s pain and illness (paragraph 62) . However, the prohibition against sale was weakening. The Latvian government’s response (see below), at footnote 19 on p.16, explains: “[t]he word smotriaschij in Russian means “the one who oversees”. They are appointed by the so-called vor v z akone (thief in law), a criminal at the peak of the criminal hierarchy, to monitor compliance with the rules of the informal prisoner hierarchy (the so-called vorovskije zakony (laws of thieves), act as intermediaries between the administration and prisoners, resolve conflicts between inmates, and manage the common fund, the so-called obschiak .”

40. The committee was concerned that the sale of drugs, closely linked to organised crime, could increase. An increase in drug testing was recommended, but not carried out by health staff as this damages therapeutic relations. At paragraph 67, the report addressed therapeutic and prevention measures. Opioid agonist therapy ( OAT ), administering methadone and buprenorphine, was the only such measure available and only at two of the four prisons visited, Daugavgrīva and Jelgave. It was not available at Jēkabpils or Valmiera.

41. The authors called for increased prevention and treatment measures including initiating OAT at all prisons, combined with education and counselling. The committee expressed concern that the Olaine addiction treatment centre was not functioning properly and was operating at under half capacity: its work was being sabotaged by the informal prisoner hierarchy because all prisoners except the levyje were banned from attending the centre and the levyje , on returning from it, were at risk of violence and sometimes had to be accommodated separately for their safety (paragraphs 68-69).

42. Of relevance to the present appeals, footnote 66 stated: “66. Reportedly, many prisoners applied to go to the Olaine Centre because it was a requirement in their individual sentence plan. However, despite applying, only a very small number, and only levyje or those ready to be demoted, actually went there, due to the informal prohibition. Allegedly, the main obstacle was the fact that, in the Olaine Centre, inmates from different castes had to keep their food in the same fridges, use the same kettles, eat at the same table etc. Smotriaschije [ an “overseer” prisoner in authority ] interviewed by the delegation claimed that they were ready to “lift the ban” if the Prison Administration ensured that prisoners from different castes could co-exist separately in the Centre as they did in prisons but, according to them, the authorities were not willing to make such a compromise.”

43. The Latvian government’s response was published, as seems to be the convention, on the same day as the 2025 CPT report, so both documents are published at the same time. Like the 2025 CPT report, it bears signs that parts of it were written some time before the publication date and that when the response was written the government of Latvia already had access to the contents of what would become the published report. The same was the case when the 2023 CPT report was published.

44. The response referred to wide ranging amendments to Latvia’s sentencing code, effective from 1 September 2024. The highlights were (most materially here) first, transition to a two tier prison system, with open and closed conditions. Prisoners can earn a transfer to open conditions after serving half their sentence. Second, prison heads can group prisoners together according to their individual needs (“individualisation of sentence execution”), which is intended to help with “resocialisation” of prisoners. The range of “resocialisation measures” is expanded in various ways.

45. The government then referred to D. v. Latvia and measures to comply with the judgment. As predicted by the prison authorities on the occasion of the CPT’s May 2024 visit, the government explained in its response that it had, on 8 October 2024, approved a report called Measures to Reduce the Informal Hierarchy in Prisons ( the October 2024 report ). The government then summarised the October 2024 report, referring to the informal hierarchy and its traditions as “a certain historical legacy which is gradually changing”; meaning, I infer, the legacy of Soviet era prisons and prison camps.

46. There was a commitment to providing infrastructure that meets the CPT’s standards and to ensuring the availability of qualified staff, increased use of technology such as video surveillance, and regular security checks. The Ministry of Justice has been instructed to report on progress by the end of 2026 and again by 1 July 2028. Regional events to discuss the judgment in D. v. Latvia took place at the end of 2024, the government records. These brought together courts, police, the probation service, prosecutors, prison authorities and Ministry of Justice civil servants.

47. As for prison occupancy and replacement of dormitories with cells, the new Liepāja Prison, under construction and due to open in 2026, would provide 1,200 of a total 1,704 places that fully meet the required standards, in low capacity cells. Of those, 200 would be places in the Addiction Treatment Centre of Olaine Prison (Prison Hospital of Latvia). Cell occupancy would, eventually, be limited to four prisoners per cell. The government noted that at the time of the CPT’s visit in May 2024 the overall prison population was 3,344 persons.

48. In response to the CPT’s call to end the use of the smotriaschije as a means of maintaining order in prisons, to segregate them from other prisoners and increase staff levels, the government commented (page 17): “Prisons continue to work purposefully on improvements, ensuring both order and safety as well as the respect for human rights within the prisons. Undeniably, prisons face staff shortages; however, the existing staff ensures order and security in prisons. Prison officials carry out the duties of monitoring prisoners, including by means of video surveillance technology, which facilitates supervision and reduces risks, such as those related to inter-prisoner violence or other illegal activities. The segregation of informal leaders and their close circle from the rest of the prison population at the lowest level is already being implemented, and this practice will be upheld in the future. In order to segregate first-time prisoners, protect them from being exposed to the informal prisoner hierarchy, and to prevent the evolution of prison subcultures, on 11 December 2024, the Administration sent out a letter to prisons instructing them to intensify control and ensure that remand prisoners who have not served a sentence in a prison before detention are allocated separately from other detained persons and prisoners … . A similar approach applies to the segregation of first-time sentenced persons. In a letter dated 28 December 2024, the Ministry of Justice informed the Administration that first-time sentenced persons must be allocated separately from the rest of prison population and that such practice should be implemented in accordance with the applicable provisions of the CESL. Pursuant to Article 131 of the CESL, when deciding on the placement of a sentenced person in a specific prison, the Head of the Administration is required to assess security and crime prevention criteria, as well as evaluate whether a first-time sentenced person is at risk of being exposed to the informal prisoner hierarchy. This obligation of assessment is, in effect, currently imposed by the ECHR judgment. Similarly, when deciding on the allocation of a sentenced person in a prison in accordance with Article 132, paragraph 1, of the CESL, their criminal background is taken into account …. .”

49. On the subject of health care and education, the government pointed to a range of meetings with health professionals and efforts to increase the availability of and standard of health care, in connection with matters such as substance addiction, infectious diseases and sexual health. On the specific subject of measures to bring the Olaine Addiction Treatment Centre back to proper functioning, the government commented that the Centre is adequately staffed and equipped. As for occupancy and the problem that attendance is de facto limited to levyje , the government responded: “The issues regarding the occupancy of the Addiction Treatment Centre, as described in the Committee’s report, have long been evaluated by the Administration, and it has been decided to implement radical changes to ensure that prisoners with addictions reach the Addiction Treatment Centre and receive appropriate help: 1) In 2025, the conditions for placing the prisoners in the Addiction Treatment Centre will be changed. In addition to the existing sections, new sections will be created for prisoners serving short-term sentences of deprivation of liberty, as the largest portion of this group have committed crimes under the influence of alcohol or drugs (e.g., drunk drivers, etc.). Those prisoners will be engaged in the improved Programme where one of the modules aims at tackling addiction-related issues. It is also planned to engage in the activities of the associations Latvian Anonymous Drug-Addicts and the Latvian Fellowship of Anonymous Alcoholics. 2) The programmes for the reduction of addictions currently implemented in the Addiction Treatment Centre will be updated within the framework of the SRRI Project (Measure 2.7 of the Act ion Plan). 3) The current model for selecting prisoners who wish to engage in the programmes offered by the Addiction Treatment Centre will be changed. Currently, the selection of sentenced prisoners for the addiction reduction programme is managed by the Addiction Treatment Centre coordinators working in prisons who consult the prisoners about the addiction reduction programme, the conditions for accommodation in the Addiction Treatment Centre, the scope and format of implementing the addiction reduction programme, as well as motivate them to participate in the programme. At the end of 2024, a pilot project was implemented in Riga Central Prison for a small group of prisoners, in which a contact person model was tested. 43 The outcomes of the pilot project have proven to be successful in several areas, and as part of the continuation of the pilot project on introducing contact persons in 2025, Riga Central Prison plans to entrust the selection of prisoners for placement in the Addiction Treatment Centre to prison officials responsible for managing the sentenced prisoner’s resocialisation case (contact persons). The good practice of the pilot project is also to be implemented in other prisons.” With the gradual introduction of a model for the system of contact persons (Activities 2.2–2.5 of the Act ion Plan), the selection of prisoners for the addiction reduction programme at the Addiction Treatment Centre in the prison will be ensured by contact persons responsible for managing the resocialisation cases of the sentenced prisoners. [ footnote: ] 43 In addition, within the framework of the pilot project, on 25 November 2024, the senior inspectors of the Resocialisation Department were trained on issues regarding the selection for the addiction reduction programme.”

50. A series of other measures was included in the government’s response, which I will not set out in detail. They covered matters such as methods to combat the entry of drugs into the prisons from outside; increased use of technology, especially video surveillance to obviate the need to recruit large numbers of staff; and resocialisation measures, training under the October 2024 action plan, to deliver training for staff and enhance their work conditions and experience. The government explained that “with the new Liepāja Prison entering into service, at least three existing prisons will be closed” (page 33). Arrangements would be made for staff transfers accordingly. Prison Conditions: Submissions

51. Both appellants have applied for permission to rely on the 2025 CPT report and the response to it, as fresh evidence not available to the judges below. The respondent accepts that this evidence was not available below and that I must look at it fully to determine whether it is decisive and therefore admissible, which the respondent disputes. The appellants say the fresh evidence would or should have led the judges below to discharge them on article 3 grounds.

52. There was no dispute about the core principles to be applied in article 3 cases. There is an absolute prohibition on inhuman or degrading treatment ( Soering v United Kingdom (1989) 11 EHRR 439 , at [88]). As to the standard of proof, before extradition can be refused there must be “strong grounds for believing that the person, if returned, faces a real risk of being subjected to torture or to inhuman or degrading treatment or punishment” ( R. (Ullah) v. Special Adjudicator [2004] 2 AC 323 , per Lord Bingham at [24]).

53. In applying that test, per Dingemans J, as he then was, in ) at [17]-[18]: Jane v Prosecutor General’s Office, Lithuania [2018] EWHC 1122 (Admin “[b]ecause of the principle of mutual trust between EU member states, membership of the Council of Europe is a highly relevant factor in deciding whether an extradited person would, in fact, be likely to suffer treatment contrary to article 3 if extradited to another member state … [the] presumption may be rebutted by clear, cogent and compelling evidence, something approaching an international consensus… For example, if there has been a pilot judgment of the European Court of Human Rights (”ECtHR”) against the requesting state identifying structural or systemic problems the presumption will be rebutted... . Where the presumption is rebutted, the burden of proof shifts to the requesting state… .”

54. As for CPT reports and comments or opinions stated in them, while the CPT is a Council of Europe body and the Strasbourg court has particular regard for its findings, the court also stated in Mursic v Croatia (2017) 65 EHRR 1 , at [113]: “As the CPT has recognised, the Court performs a conceptually different role to the one assigned to the CPT whose responsibility does not entail pronouncing on whether a certain situation amounts to inhuman or degrading treatment … within the meaning of Article 3 … . The thrust of the CPT activity is pre-emptive action aimed at prevention which, by its very nature, aims at a degree of protection that is greater than that upheld by the Court when deciding conditions of detention.”

55. I do not need to say more about the Strasbourg jurisprudence on article 3 in prison conditions cases because I have already, in the factual context of this case, cited extensively from the decision in D. v. Latvia in which the relevant previous authorities are discussed. On the domestic front, I have already mentioned the decision of the Divisional Court in Danfelds and of Swift J in Vascenkovs . The submission of the appellants is that developments since those cases were decided and indeed since the judgments below have now rebutted the presumption of compliance in the cases of these two appellants.

56. As you would expect, the appellants’ submissions focus on the negative points in the 2023 and 2025 CPT reports and the frustration expressed in the latter at the lack of progress in the two years between the visits in May 2022 and May 2024. Informal hierarchies remain entrenched, fuelling inter-prisoner violence which usually goes unpunished and is not investigated. Accommodation in dormitories rather than cells remains prevalent. The authorities rely on the smotriaschije to keep order because of staff shortages.

57. The lot of the levyje has barely improved and amounts to degrading treatment. While drug consumption in the prisons is on the rise, those seeking drug rehabilitation treatment at the Olaine Addiction Treatment Centre are either already levyje or become one as a consequence of obtaining the treatment. Limited opioid agonist therapy is the only treatment available and only in two prisons. Addict prisoners who are not levyje are denied the chance of early release unless they attend for treatment at Olaine and thereby become levyje .

58. The appellants criticise the government’s response as inadequate. It states aspirations rather than achievements and relies too much on processes (i.e. structural and legislative reform) rather than concrete steps taken. The measures being taken, while commendable, are incapable of dismantling the enduring historical legacy. The response does not, as it should, attest to a rapid reduction in use of dormitories, nor to segregation of informal leaders from those they lead, nor to substantial increased recruitment and filling of staff vacancies.

59. On the strength of that fresh evidence, the appellants make three submissions. First, they say the 2025 CPT report provides objective, reliable, specific and properly updated evidence of systemic deficiencies which affect certain groups of people, particularly those belonging to the lowest caste of prisoners, including drug addicts who owe debts inside and outside prison. That is the position of Mr Igoniņs. DJ Clarke was wrong, Mr Hyman submits, to dismiss the dilemma he faces between foregoing treatment, retaining his caste status but losing the chance of early release, or receiving treatment and becoming an untouchable.

60. Second, D. v. Latvia is authority that the persistence of the informal hierarchy is inhuman and degrading treatment, even without physical violence. Although not expressed as a “pilot judgment” (which is not indispensable to rebutting the presumption), the court in D confirmed that the treatment that violates article 3 can be a person’s status rather than specific incidents. Latvia is obliged to take systemic measures to address the judgment. DJ Clarke placed too much weight on Vascenkovs , which is now overtaken anyway.

61. Third, the appellants submit, the Latvian government’s response to the 2025 CPT report shows that it is unwilling or unable to take the necessary measures to protect those like Messrs Sostaks and Igoniņs. The failure to take measures to protect “those in the lowest stratum in the hierarchy” (as it is put in the appellants’ joint skeleton argument), while instead proposing measures that are “delayed, partial and likely to be ineffective”, must impel the court to the conclusion that Latvia has lost the presumption of compliance with article 3.

62. Alternatively, the appellants submit, the court should seek further information from the Latvian authorities about the concrete measures the government will take to protect the appellants; or the court should seek an appropriate assurance. The court has a discretion to do so and should exercise it rather than return these appellants to Latvia. The appellants would wish to be heard further on the terms of any request for further information or an assurance.

63. The respondent advocated a much more upbeat view of the progress being made to improve conditions and disband the informal hierarchies in Latvia’s prisons. Prison conditions in a particular country are unlikely to be static and the domestic court must proceed on the most up to date evidence available: Elashmawy v. Court of Brescia, Italy [2015] EWHC 28 (Admin) per Aikens LJ (judgment of the court) at [90]. Thus, the respondent submitted, D. v. Latvia is of no real assistance: the treatment found to breach article 3 occurred between eight and 17 years ago.

64. The 2025 CPT report showed, said the respondent, that inter-prisoner violence had not deteriorated since the CPT’s visit in May 2022. A deterioration would be required to justify revisiting the decisions in Danfelds and Vascenkovs . The CPT’s stance in the 2023 CPT report, considered in Vascenkovs , was that levyje were being subjected to treatment in violation of article 3. The same conclusion is reached in the 2025 CPT report. The committee said (paragraph 59) that it “reiterates” that view; the situation of the lowest caste “could be considered … a continuing violation of Article 3”.

65. Moreover, said the respondent, there have been improvements: the rules of the informal hierarchy were becoming vaguer and the pressure to accept them less intense. Construction of cellular accommodation is not an aspiration; it is taking place now. The new block with double occupancy cells at Valmiera Prison is close to coming on stream if it is not already receiving prisoners. The 1,200 capacity cellular block prison at the new Liepāja Prison is due to open in 2026. With a prison population of under 4,000 that is significant progress. The CPT delegation was told in May 2024 that the new block was then due to open in the summer of 2024; see paragraph 20 of the 2025 CPT report. In fact it began to be commissioned in January 2025 and as at February 2025 was expected to start receiving prisoners in March 2025 in accordance with the action plan (see page 8 of the government’s response).

66. Other significant measures since the CPT’s visit in May 2024 and the Strasbourg court’s decision in D. v. Latvia are explained in the government’s response: changes to sentencing rules allowing greater sentencing progression, resocialisation measures and the grouping of prisoners according to need. Further the segregation of the smotriaschije is “already being implemented” and “will be upheld” (page 17 of the government’s response).

67. Further steps are being taken, said the respondent, to recruit additional staff and to adjust the criteria for receiving drug addiction treatment, to ensure the right prisoners receive it. The picture overall is not one of decline but of historically slow and now steady progress and improvement. The appellants would be extradited to conditions better than those at the time of Vascenkovs ; the presumption of compliance was not rebutted then; a fortiori , it cannot be now.

68. The respondent submitted that neither appellant runs the risk, if extradited, of becoming a member of the lowest caste. Mr Sostaks did not refer in evidence to having been one when last imprisoned in Latvia. Mr Igoniņs confirmed that he was not; he was a mužik or middle-tier prisoner. Both express concern at the prospect of becoming an untouchable if they receive drug treatment but, as DJ Clarke correctly observed, having to serve a sentence in full is not a violation of article 3. Both appellants can point to treatment received at HMP Wandsworth and both stand to benefit from improved pathways to treatment in Latvian prisons and inroads into the power of the informal hierarchy. Prison Conditions: Reasoning and Conclusion

69. Mindful of the strength of the presumption of compliance and the cogency of the evidence required to rebut it, I start by examining the foreseeable consequences of sending the person to the receiving state, bearing in mind the general situation and the personal circumstances of the person. I begin with the evidence about prison conditions in Latvia at a general and systemic level. In my judgment the evidence establishes that the informal prisoner hierarchy is persistent and long standing. However, pressure on Latvia from the CPT and the Strasbourg court is beginning to yield results.

70. The pressure is to bring the informal hierarchy to an end, break the power of the informal leaders, i.e. the smotriaschije and the blatnyje , and end the degrading treatment of the lowest caste members, the untouchables, kreisie or levyje . The culture is starting to change, partly by natural evolution and partly because the measures alluded to in the latest government response are having an effect. The change is slow; there is much still to do. It is likely that the situation of the worst off prisoners will slowly improve, relative to the rest.

71. I accept the authority of Vascenkovs for the proposition that the Latvian prison regime was generally article 3 compliant when that case was decided and that the presumption of compliance was not at that stage rebutted, notwithstanding the 2023 CPT report which was considered by Swift J, along with the government’s response to it. Vascenkovs was not a kreisie ( levyje ) case; the caste system was not relied on, though inter-prisoner violence was considered. Nor was it a case involving drug addiction and treatment. The warrant was an accusation warrant in respect of alleged benefit fraud.

72. At that time, there was already case law in the European Court of Human Rights establishing that an individual had suffered treatment in violation of article 3 arising from conditions in detention (viz. the alleged police informer in DF v. Latvia ). That finding in itself came nowhere near rebutting the presumption of compliance for extradition purposes, as the domestic case law shows. The same is true of the finding in D. v. Latvia that the sex offender imprisoned from 2008 to 2017 suffered degrading treatment in violation of article 3. That finding related to that individual at the time of his incarceration.

73. The respondent is right to point out that D. v. Latvia is not the first Strasbourg case to address the caste system and not the first case to recognise that degrading treatment in violation of article 3 can arise from a person’s prison status and subjection to demeaning tasks; it need not involve actual physical violence or an imminent threat thereof. As the case law shows, evidence of a breach of article 3 in one case at an earlier time is not necessarily cogent and convincing evidence of the necessary risk of breach in another case at another, later time.

74. That said, the Strasbourg court in D. v. Latvia assumed a role beyond examining the narrow issue of D’s treatment up to 2017. It might be thought logically unnecessary to mention the 2023 CPT report which dealt with conditions as at May 2022, more than four years after D’s complaint was filed. I suppose conditions in May 2022 could in principle throw some retrospective light on conditions earlier, up to 2017. But the primary reason for the court’s extensive post-2017 findings was to exert pressure for change in Latvian jails.

75. Against that background, I do not accept the appellants’ submission that Latvia has lost the presumption of compliance with article 3 because the measures taken in the last year and a half are inadequate to protect against the risk of degrading treatment of prisoners generally, in breach of article 3. At the most, it could be said that a person (for example, a serious sex offender) who is virtually certain to become an untouchable, could be at risk of such treatment.

76. I do not find it necessary to decide in these appeals whether that is the position, or whether the measures and the pathway to improvements since May 2024, explained in the response to the 2025 CPT report, are enough to negate such a risk. If it were necessary to address the issue head on, I might have decided to seek an assurance or further information from the Latvian authorities on what measures would be taken to protect the person against degrading treatment.

77. I come next to the position of the two appellants as individuals. I accept the respondent’s submission that Mr Sostaks did not in his evidence rely on any argument that he would necessarily be treated as an untouchable outcast if he is extradited to Latvia. His evidence was that he had drug debts. His counsel, Mr Meredith, rightly conceded in oral argument that owing debts (or drug related debts) alone did not inexorably lead to being treated as an untouchable.

78. I reject the submission that the jeopardy of becoming an untouchable through receipt of drug rehabilitation treatment is anywhere near sufficient to rebut the presumption of compliance with article 3. I find that it is not; first, because of the improving situation in Latvian prisons and in drug addiction treatment; and second, because I agree with DJ Clarke’s proposition in Mr Igoniņs’ case that having to serve the whole of a sentence is not degrading treatment in violation of article 3. I therefore reject this ground of appeal in Mr Sostaks’ case.

79. The same analysis applies to Mr Igoniņs. When last imprisoned in Latvia he was a mužik . He said in evidence that he was a police informer, addicted to fentanyl and in debt to drug dealers. He might well become an untouchable if he chose to attend for drug rehabilitation treatment, in the hope of early release. But that does not come near to rebutting the presumption of compliance. At worst, he would have to serve his sentence in full, probably again as a mužik , his status when last incarcerated. That is not degrading treatment. I therefore reject Mr Igoniņs’ only ground of appeal and his appeal must fail. Mr Sostaks’ Appeal: Submissions on Proportionality

80. Mr Sostaks’ additional ground is that his extradition would be disproportionate under section 21 A(1)(b) of the 2003 Act . In summary, he says first that he has already served about 21 months on remand in custody at HMP Wandsworth; second, that having regard to the seriousness of the crimes of theft of which he stands accused and the likely penalty if he were convicted, a custodial sentence would be highly unlikely if he were extradited; and third, if he were convicted and sentenced to custody, the 21 months served here would be deductible and it is unreal to suppose any custodial sentence in Latvia would exceed 21 months.

81. By section 21 A of the 2003 Act the court is required to: “(1) … decide … (a) whether the extradition would be compatible with the Convention rights within the meaning of the Human Rights Act 1998 ; (b) whether the extradition would be disproportionate. (2) In deciding whether the extradition would be disproportionate, the judge must take into account the specified matters relating to proportionality (so far as the judge thinks it appropriate to do so); but the judge must not take any other matters into account. (3) These are the specified matters relating to proportionality— (a) the seriousness of the conduct alleged to constitute the extradition offence; (b) the likely penalty that would be imposed if D was found guilty of the extradition offence; (c) the possibility of the relevant foreign authorities taking measures that would be less coercive than the extradition of D [ the requested person ].”

82. In written argument, Mr Meredith submitted that the authorities on section 21 A pointed to extradition being disproportionate here. He referred me to ( inter alia ) the guidance of this court (Pitchford LJ and Collins J) in Miraszewski v. District Court in Torun, Poland [2015] 1 WLR 3929 ; the Lord Chief Justice’s Guidance on assessment of serious of conduct alleged in extradition cases, now at paragraphs 12.2.2-12.2.4 of the Criminal Practice Directions 2023; Kozar v District Court in Liberec, Czech Republic [2024] EWHC 2226 (Admin) (Julian Knowles J); and TACA article 597.

83. The latter refers to cooperation through the arrest warrant being “necessary and proportionate” and mirrors the language of section 21 A of the 2003 Act , referring to “the seriousness of the act, the likely penalty that would be imposed and the possibility of a State taking measures less coercive than the surrender of the requested person particularly with a view to avoiding unnecessarily long periods of pre-trial detention”. Mr Meredith reminded me that in Vascenkovs Swift J had rejected the suggestion that the advent of the TACA regime after Brexit had altered the approach to section 21 A ordained in Miraszewski .

84. While accepting that Mr Sostaks had previous convictions in Latvia and there were five offences, Mr Meredith submitted that they were theft offences and relatively trivial; they involved shoplifting and theft of bicycles. They fell squarely within the Lord Chief Justice’s guideline and represented the kind of offending the proportionality bar was intended to weed out. The district judge below had failed to appreciate this and had placed too much weight on Mr Sostaks’ previous convictions.

85. The maximum penalties of one and two years’ imprisonment were not of much assistance, Mr Meredith submitted. Latvia had provided no information about the likely sentence in the event of convictions. The judge had been wrong to decide that the offending could well attract a custodial sentence in this country. The most likely penalty here for the theft of the two bicycles would be a low level community order within category 3B of the Sentencing Council’s guideline on theft. For the other thefts, the penalty would probably be a fine or even a discharge, in category C3 of the guideline on theft from a shop or a stall.

86. Mr Meredith submitted further that any prison sentence in Latvia might well be suspended and, finally and decisively, even if it were not, any term of immediate imprisonment could not exceed the time already served here on remand in custody, now 21 months. At the time of the judge’s decision below, it had been about 8 months. Latvia would be obliged by TACA article 624 to deduct that time from any immediate sentence of imprisonment. It is inconceivable, said Mr Meredith, that Mr Sostaks would be sentenced to upwards of 21 months’ immediate imprisonment, if he were extradited, convicted and sentenced.

87. In her written skeleton, Ms Bostock for the respondent took what amounts to a preliminary point of law, though not described as such. She was responding to a contention in the perfected grounds of appeal that the judge below having “noted the period … served on remand in custody awaiting the final hearing”, erred in finding that “the allegations made could attract a prison sentence if the Applicant was convicted in this jurisdiction”.

88. Ms Bostock countered that in her skeleton, saying any period spent on remand in this country awaiting extradition “is not one of the specified matters relating to proportionality which a judge is entitled to take into account under this heading”. That was because, she explained in oral argument, the list of permissible relevant considerations in section 21 A(3) was exhaustive: by subsection (2) “the judge must not take any other matters into account”.

89. The respondent, therefore, is saying that the “likely penalty that would be imposed” by the requesting state’s courts must be calculated by a judge applying section 21 A(3)(b) without reference to any deduction by the requesting state of any time spent on remand in custody in this country, awaiting extradition.

90. I asked Ms Bostock whether she accepted that at least two judges of this court, Dove J (as he then was) and Julian Knowles J respectively had adopted the contrary interpretation of “likely penalty” when ordering the discharge of the appellant in each of Kozak v Buda District Court, Hungary [2023] EWHC 149 (Admin) , (see at [10]ff), and Kozar v District Court in Liberec, Czech Republic [2024] EWHC 2226 (Admin) , (see at [61]-[63]).

91. In each case, time spent on remand was an important element of the “likely penalty” assessment. Ms Bostock’s response was that, respectfully, both judges were wrong to take that factor into account; it was off limits because of the words at the end of subsection (2). I asked why they were wrong and it became clear that the novel point of statutory construction needed to be properly aired in further written submissions, which I directed. Counsel and I were concerned that there could be other relevant authorities not before the court.

92. In further written argument, Ms Bostock submitted that Dove J’s attention appeared not to have been drawn to the guidance of the Divisional Court in Miraszewski at [37] and [39], where Pitchford LJ states that the “likely penalty” consideration is focused “upon the likelihood of a custodial penalty” as opposed to the length thereof; while recognising that non-custodial penalties can also justify extradition because the court should “recognise and give effect to the public interest in prosecution”.

93. She also cited Fordham J’s permission decision in a conviction warrant case, Molik v Judicial Authority of Poland [2020] EWHC 2836 (Admin) ), saying it showed the correct route for a “time served” argument is the first limb of s.21 A(1)(a) (whether extradition would be compatible with Convention rights) not the second (whether extradition would be proportionate). At [17], Fordham J made clear that where time served on remand exceeds time to serve, extradition is necessarily inappropriate, whether as an abuse of process or via article 8 rights.

94. In Molik , the applicant had six weeks of a 10 month activated custodial sentence to serve when permission was refused at the oral permission hearing, Ms Bostock pointed out. Fordham J had cited Aikens LJ’s judgment in Kloska v. Circuit Court of Katowice, Poland [2011] EWHC 1647, at [27] for the proposition that “except in most unusual circumstances, it cannot be for the courts in England to form a view on whether the person to be extradited has or has not served enough of his sentence that was imposed by the requesting judicial authority”.

95. Fordham J maintained the decision to extradite even though it was clear that by the time of any substantive hearing the line would have been crossed and the applicant would have served more time on remand than his outstanding sentence in Poland and would be entitled to discharge. Ms Bostock submitted that where a person had yet to be sentenced, the public interest in extradition was even stronger: it was for the requesting state’s courts to determine sentence and the English court should not circumvent that principle by recourse to an interpretation of “likely penalty” that included time spent on remand here.

96. Ms Bostock further submitted, as she had orally, that the rationale for the section 21 A(3) factors was, as expounded in Miraszewski , that they focussed on low level offending that would not be met with any custodial sentence at all. If the focus was to be on the likely length of the custodial sentence, that should be argued as a matter of Convention rights. The seriousness of the conduct or alleged conduct of the requested person is an objective constant. It does not change according to how much time has been served here on remand.

97. She submitted that any Convention rights issue must be determined first because “the statute must be followed in order”. For that proposition she cited Raffile v USA [2004] EWHC 2913 (Admin) (Rose LJ and Pitchford J as he then was). The relevant passage appears to be at [33], where Pitchford J said in a different context that the procedure before the judge below “was a staged progress through the relevant sections, each of which was dependent on a decision previously made”; and that the judge: “could not be expected to reopen those decisions at such a late stage of the hearing. In fact, passage of time under section 79(1)(c) was three steps back in the procedure. Two of those steps involved decisions made under section 84 and 85 which were uncontested; while, on the other hand, section 79, to which the district judge's attention was being redirected, created statutory bars to extradition.”

98. Aside from her point of statutory interpretation, Ms Bostock submitted that even if her interpretation was wrong and that of Dove J and Julian Knowles J was right, Ms Sostaks’ appeal should be dismissed and he should be extradited to Latvia to face trial for the five thefts. In support of that submission, she made the following further points.

99. It was not a bar to extradition per se that the requested person would be likely to receive a non-custodial sentence: Miraszewski , at [39]). The purpose of the proportionality bar went no further than that of providing a further brake on extradition for very minor offences, over and above the requirement that the conduct alleged must be punishable by at least 12 months’ detention ( section 64(3) (c) and (4)(c) of the 2003 Act ); per Swift J in Vascenkovs at [11].

100. A comparison with sentencing practice in this country only gives a general idea of the likely sentence as no trial has taken place and the full facts are unknown ( ibid. at [24]). Considerable latitude must be accorded to the issuing state. The appellant, illegitimately, seeks to use the proportionality bar in section 21 A(1)(b) to argue that he has in effect already served his sentence. That overlooks the public interest in extradition and undermines mutual trust.

101. If Mr Sostaks is not extradited, Ms Bostock said, he will not face trial at all for the five offences and they will not appear in his record of convictions in Latvia. That tends to undermine the international comity emphasised in Andrysiewicz v. Circuit Court in Lodz, Poland [2025] 1 WLR 2733 , per Lord Lloyd-Jones and Lord Stephens JJSC at [62]. The requesting state is prevented from managing the offender’s punishment for the crimes for which he is never tried.

102. Finally, the respondent asserts that on the facts the offending is far from trivial. There are five distinct offences of theft; four carry a maximum sentence of one year’s detention, the fifth, of two years’ detention. The appellant has a bad record in Latvia and in this country. Multiple offences and previous convictions are matters singled out as possible exceptions in the Lord Chief Justice’s guidelines on low level offending, which may make discharge inappropriate.

103. Mr Meredith, for Mr Sostaks, also responded in writing to my invitation to address the respondent’s interpretation of section 21 A(3)(b) and the meaning of the words “likely penalty that would be imposed” in that paragraph. He submitted, straightforwardly, that Dove J in Kozak and Julian Knowles J in Kozar were both correct to include time served on remand in their respective proportionality assessments.

104. He also disputed the respondent’s proposition that the issue of time served on remand must be addressed through the lens of Convention rights and not proportionality simpliciter . He pointed out during the hearing that the same factual points may be deployed in respect of different legal issues, where the factual material overlaps. Thus, it is very common for a requested person to rely on the same facts to support a passage of time argument (under section 14 of the 2003 Act ) and an article 8 argument (under section 21 or section 21 A).

105. It is a matter for the court in what order it addresses the issues, Mr Meredith submitted. Thus, in Kalinauskas v. Prosecutor General’s Office, Lithuania [2020] EWHC 191 (Admin) , the Divisional Court (Irwin LJ and Supperstone J) considered it unnecessary to determine the appellant’s articles 3 and 8 grounds of appeal, having allowed the appeal under section 21 A(1)(b) of the 2003 Act per Supperstone J at [9]-[10]; at [20] being in “no doubt that if the Appellant was to be sentenced now for the offending, he would be immediately released”; and noting at [21] that “he has now served in excess of any sentence that could have been imposed for his conduct… .”

106. Mr Meredith added that a similar approach was adopted by Chamberlain J in Buivis v. Deputy Prosecutor General, Lithuania [2021] EWHC 2042 (Admin) , in which he followed the reasoning in Kalinauskas to the effect that deductible time served on remand is relevant to the “likely penalty” issue; and by Holman J in Lucki v Poland [2022] EWHC 818 (Admin) , adopting the same approach. In both cases, the court was referred to and aware of the Divisional Court’s guidance in Miraszewski .

107. As for the permission decision in Molik , Mr Meredith submitted that Fordham J recognised that extradition should not occur where time served here exceeds time to be served in the requesting state. While he framed that proposition in terms of abuse of process or article 8 rights, he did not exclude the same proposition viewed as a section 21 A(1)(b) issue. Moreover, the same judge granted Mr Sostaks permission to advance the proportionality ground of appeal in this case, reasoning that they had arguable merit alongside article 8 as “a possible wider alternative”. Mr Sostaks’ Appeal: Reasoning and Conclusions on Proportionality

108. As already noted, the language of section 21 A is now reflected in TACA article 597. Both refer to the triple concepts in subsection (3)(a)-(c): the seriousness of the act, the likely penalty that would be imposed and the possibility of a state taking less coercive measures. In my judgment, it is significant that these three concepts are chosen in article 597 “particularly with a view to avoiding unnecessarily long periods of pre-trial detention”.

109. TACA article 624 is the provision for deduction of time served in detention in the executing state: “ Deduction of the period of detention served in the executing State

1. The issuing State shall deduct all periods of detention arising from the execution of an arrest warrant from the total period of detention to be served in the issuing State as a result of a custodial sentence or detention order being passed.

2. All information concerning the duration of the detention of the requested person on the basis of the arrest warrant shall be transmitted by the executing judicial authority or the central authority designated under Article 605 to the issuing judicial authority at the time of the surrender.”

110. That provision does, I acknowledge, envisage that the court imposing a custodial sentence on an extradited person after surrender will state the period of detention without any deduction; and that the amount notified on surrender will then be deducted, perhaps administratively, from the period of detention decided upon by the sentencing judge. However, that judge could instead sentence the extradited person, after surrender, to a shortened period of detention which already includes the necessary deduction.

111. The latter course would not be objectionable. It should be a matter for the law of the receiving (issuing) state to determine whether the deduction is made at the point of sentencing or whether it is stated to “count towards” the period of detention (as happens here where time is spent on pre-sentence remand in custody or on tagged curfew). Indeed, the law of the receiving state may leave the matter to the sentencing judge. Any of these methods would satisfy the deduction requirement in article 624.

112. These observations incline me to a flexible approach to the words “likely penalty”. The guidance of the Divisional Court in Miraszewski does not compel a more rigid approach. The issue of statutory construction I am addressing was not considered in Miraszewski . The judges who have subsequently opted for the flexible approach advocated by Mr Sostaks were, for the most part, made aware of the Divisional Court’s guidance, if they were not already.

113. The permission decision in Molik is not directly in point. It was a conviction warrant case where the amount of time that remained to be served (six weeks) was known. Again, the issue before me did not arise for decision. If it had, it is likely that Fordham J would also have adopted the flexible interpretation, which is implicit in the order he made granting permission to Mr Sostaks to advance this ground in this appeal. He specifically mentioned the time served.

114. I acknowledge Ms Bostock’s arguments that the court in Miraszewski emphasised the importance of whether a sentence in the issuing state would be custodial rather than how long any custodial sentence would be; and the importance of mutual trust and international comity. But it is sometimes impossible, on either interpretation of “likely penalty” to avoid paying any attention at all to the likely length of a custodial sentence in the issuing state.

115. Even on Ms Bostock’s interpretation, some regard may be had, in general terms, to what the period of a custodial sentence would be in this country. It is clear from Miraszewski and other cases that this is so. The question is not just the binary one whether the sentence in the issuing state would be custodial or not. Just as a person may be extradited even if a custodial sentence in the issuing state is not likely, a person may be discharged in an appropriate case even if a custodial sentence in the issuing state is likely.

116. An examination of sentencing practice in this country, though no more than a rough guide, is legitimate particularly if there is no material from the requesting state to guide the court in this country, apart from the maximum sentence for the offence or offences. It is open to the Crown Prosecution Service or National Crime Agency to ask for an indication of the likely sentence range in an accusation warrant case such as this, while recognising that the issuing state may be reluctant to commit itself before any trial has taken place.

117. I reject the respondent’s submission that the issues that arise under section 21 A must be followed in the order in which they appear, i.e. Convention rights first and proportionality afterwards. I do not accept that Raffile v. USA is authority for that proposition; the context was wholly different. The submission is inconsistent with the course followed by the Divisional Court in Kalinauskas at [10] and by Julian Knowles J in Kozar at [68].

118. The decision in Kalinauskas directly supports Mr Sostaks’ flexible interpretation of “likely penalty that would be imposed”. So does Holman J’s decision in Lucki and Chamberlain J’s in Buivis . In the latter case, the question I am considering actually arose. Chamberlain J said this at [19]: “For the Lithuanian authority, Ms Brown submitted that the only question under s. 21 A(3)(b) was ‘the likely penalty that would be imposed’. This meant that the question whether that penalty exceeded the time already spent on remand was not, strictly, a matter for consideration under s. 21 A at all.”

119. At [21], he added: “(c) The principal focus of s. 21 A(3)(b) (‘the likely penalty that would be imposed’) is on the proportionality of extraditing someone who is not likely to receive a custodial sentence in the requesting state: Miraszewski , [37]. (d) Looking purely at the language of s. 21 A(3)(b), particularly taken with s. 21 A(2), there is some force in the respondent’s submission that the court should focus narrowly on the penalty actually imposed, rather than on whether there will be time left to serve after that spent on remand in the executing state is deducted. However, there is an argument to the contrary based on a purposive reading of s. 21 A(3)(b): there is a diminished public interest in extraditing someone who is not likely to serve any time in custody in the requesting state. The Divisional Court’s decision in Kalinauskas is only consistent with the view that time spent on remand is relevant. This is not clearly wrong. I must therefore follow it.”

120. I respectfully agree and I consider the latter interpretation to be the correct one. It is supported by the weight of existing authority. The better view is that “likely penalty” can include time served in the executing state. It is not contradicted by any authority directly on the point. It accords with the policy of discouraging long periods of pre-trial detention, mentioned in TACA article 597. It avoids arbitrary consequences which may depend on how an issuing state’s sentencing laws are framed or on the discretion of an individual sentencing judge.

121. The rigid interpretation of “likely penalty” contended for by the respondent could, in an extreme case, work serious injustice if, say, a requested person were to spend three or four years on remand here and then be sent to an issuing state to be tried for an offence worth, at the outside, say one year’s imprisonment. The more merciful flexible interpretation is, fortunately, not linguistically untenable and much to be preferred. It need not do violence to mutual trust or international comity: our courts can best achieve extradition in short sentence accusation cases by accelerating the process or granting bail, or both.

122. It remains to consider the application of section 21 A(1)(b) to the facts here, taking account of the 21 months served. Four of the five alleged offences of theft are relatively minor. The first is theft in March 2021 of six packs of batteries valued at about €60. The second is theft of eau de toilette worth €63, six days later. The maximum sentence is one year in each case.

123. The third offence is more serious: in June 2021, stealing two bicycles together valued at €710, from “the staircase” of “a multi-apartment building”. That carries a maximum sentence of two years. The fourth and fifth are minor like the first and second: theft in March 2021 of four bottles of vodka, together worth about €136; and earlier, in February 2021, of four bottles of whisky and two of rum, together valued at €181. The total value of the goods allegedly stolen is therefore €1,014. There is multiple offending.

124. And Mr Sostaks has previous convictions for similar offences, here and in Latvia. Here, he has at least two convictions for thefts committed in 2023, said to be drug related. In Latvia, he appears to have six convictions in the period from February 2021 to April 2022. All are for theft or kindred offences. The sentences for the minor thefts were community service or 40 days’ imprisonment. The two worst we would call burglaries. For those offences, the sentences were five months’ and one year’s imprisonment respectively.

125. After anxious deliberation, while the case is close to the borderline I have come to the clear conclusion that extradition of Mr Sostaks would not be disproportionate. I accept that most of the offences are minor and that he has already served 21 months on remand here. But he is a recidivist thief and burglar for whom community sentences in Latvia have not worked.

126. The sentencing regime in this country is far less relevant than the sentences he has already received from the courts of Latvia. In this country he might well be sentenced as a “third strike” burglar if convicted of the theft of the bicycles from the stairs of an apartment block, if the stairs were inside the building. But that is by the by. His longest sentence so far in Latvia is one year and the pattern is, as you would expect, one of increasingly severe sentences.

127. It is not my prediction that Mr Sostaks will necessarily receive, if convicted, a custodial sentence, once the 21 months served here has been deducted. But that is not conclusive against the proportionality of extradition. Where a case is close to the borderline, the executing state’s court should be slow to find extradition disproportionate; it is better to defer to the issuing state’s justice system. It is their case, not ours. I also accept Ms Bostock’s submission that there is a public interest in prosecution of Mr Sostaks in Latvia, whatever the sentence. If he is guilty, the convictions should be added to his criminal record. Disposal

128. I am grateful to counsel for their helpful submissions. I dismiss the appeal of Mr Igoniņs on the article 3 ground. I dismiss the appeal of Mr Sostaks on both the article 3 and proportionality grounds. I hope that future cases where time served on remand is in issue will reduce. The problem can be addressed by efforts to speed up the extradition process in this country, if granting bail is not appropriate; especially in accusation warrant cases where a short if any custodial sentence is likely, should there be a conviction after extradition.

Janis Sostaks & Anor v Prosecutor General’s Office, Latvia [2025] EWHC ADMIN 2795 — UK case law · My AI Travel