Pandemic lockdown restrictions and COVID-19 hospitalization and deaths in patients with rheumatic and musculoskeletal diseases

doi:10.21203/rs.3.rs-1584534/v1

Background: A wide range of response strategies have been implemented by different countries to slow the spread of the COVID-19 outbreak. While the effects of these strategies in the general population have been investigated, less is known about the effects of these strategies on those with pre-existing comorbid conditions including rheumatic and musculoskeletal diseases (RMDs). We aimed to quantify the impact of stringent lockdown restrictions in Norway compared to Sweden on COVID-19 hospitalization/death and all-cause mortality in persons with and without RMDs.

Methods: Using ICD codes from register data, individuals aged ≥35 years residing in Norway and Skåne region (Sweden) on 31^st December 2019 were divided into the RMD cohort (those with any diagnosis of RMDs during January 1, 2017–January 31, 2020), and the non-RMDs cohort. We obtained daily data on the outcomes for February–May, 2020 and applied difference-in-difference and event study to estimate the causal effect of lockdown.

Results: Compared with Skåne, strict lockdown in Norway was, on average, associated with 18.0 (95% CI 8.6, 27.3), 2.7 (1.5, 3.9), and 2.9 (–4.0, 9.7) fewer daily COVID-19 hospitalizations, COVID-19 death, and all-cause mortality per million persons, respectively, among individuals with RMD. Corresponding reductions were 9.1 (95% CI 4.5, 13.7), 0.9 (0.4, 1.5), and 2.9 (–3.8, 9.7) among those without RMD. Event study showed that the effects on COVID-19 hospitalization and death were evident from 1-2 weeks after lockdown implementation and onward with generally greater effects in the RMDs than the non-RMDs cohort and in older than younger individuals.

Conclusion: Strict lockdown was associated with reductions in COVID-19 hospitalizations and (to a lesser extent) COVID-19 death with no essential effect on all-cause mortality. Our findings confirm effect of strict lockdown in particular among those at higher risk of COVID-19 adverse outcomes such as older persons with RMDs.

COVID-19

Difference-in-difference

Event study

Lockdown

Norway

Sweden

A wide range of response strategies have been implemented by different countries to slow the spread of the COVID-19 outbreak [1]. For instance, Sweden and Norway, two Scandinavian countries with significant economic, cultural and institutional similarities, adopted substantially different response strategies [2-4]. While in March 2020, Norway government enforced a five-week national lockdown, Sweden government adopted a lenient mitigation strategy that mainly focused on recommendations to take voluntary actions [5]. To aid informed policy-making, several studies used the data from these two countries to assess the impact of their divergent restriction strategies on health and economic outcomes in the general population [2-4, 6, 7]. However, little is known about the effects of these strategies on subpopulation with potentially higher risk of more severe COVID-19 outcomes including those with pre-existing comorbid conditions [8, 9]. One such subpopulations are patients with rheumatic and musculoskeletal diseases (RMDs) who are among subgroups with high rates of comorbidities [10, 11] and recent studies reported higher risk of COVID-19 and COVID-19 hospitalization or death for persons with than without RMDs [12-14]. In addition, a recent study in Sweden reported greater reductions in healthcare use following COVID-19 among people with than without RMDs [15]. This study exploited the variation in response strategies between Norway and Sweden to quantify the impact of lockdown restrictions on persons with and without RMDs.

We used register data from the population and health register covering all inpatient healthcare in Skåne, Sweden and the Norwegian Emergency Preparedness Register (BEREDT C19) in Norway. Skåne is the southernmost region of Sweden with about 1.4 million inhabitants (13% of the Sweden population). The Skåne Healthcare Register is a regional legislative administrative healthcare database containing the diagnostic codes from publicly funded clinics according to the International Classification of Diseases 10 (ICD-10) system assigned by physicians at the time of the healthcare consultation. The BEREDT C19 register is a recently established register covering the whole Norwegian population to provide the necessary knowledge about the spread COVID-19 and its effects on population (https://www.fhi.no). It contains data from the Norwegian Surveillance System for Communicable Diseases, the Norwegian Patient Registry and the Norwegian Intensive Care Unit and Pandemic Registry. The study was approved by the Ethical Review Board, Sweden and the Regional Ethics Committee South-East, Norway. The need for individual informed consent was waived by the Ethics committees (i.e. the Ethical Review Board, Sweden and the Regional Ethics Committee South-East, Norway). The study was performed in accordance with the Declaration of Helsinki.

Study cohorts

From these registers, we identified people aged ≥35 years who were resident in Skåne, Sweden or Norway on 31 January 2020. Among these, we defined two cohorts: the RMDs cohort as those with ≥1 RMDs diagnosis (ICD-10 codes M00–M99, L40.5) between January 1, 2017 and January 31, 2020, and the non-RMDs cohort as those with no RMDs diagnosis during this period. We also defined participants with osteoarthritis (ICD-10 codes M15–M19) among the RMDs cohort.

Outcomes

We studied three outcomes: COVID-19 hospitalization (ICD-10 codes U07.1–U07.2), COVID-19 death (defined as death within 30 days from a CODIV-19 hospital admission), and all-cause mortality. From the registers, we obtained daily counts of these outcomes from February 15, 2020 to May 31, 2020.

Analysis

To estimate the effect of strict lockdown in Norway on the outcome of interest, we applied difference-in-difference (DD) and event study designs [16] treating Sweden as control and Norway as intervention units. While DD provides an average effect for post-lockdown periods, event study provides an estimate for each period accounting for potential heterogeneity. Moreover, the validity of DD design relies on the parallel trend assumption and event study can be used to assess the plausibility of this assumption [16]. National lockdown in Norway were initiated at March 12, 2020 and to account for lag between policy implementation and its impact, we considered first two weeks (i.e. March 13–March 25) as pre-lockdown period. We divided our data in 7-day periods and included daily data from four-week prior (February 27–March 25, 2020) to nine-week after (March 26–May 27, 2020) the lockdown.

We then used DD and event study with daily rates as outcome variable and estimated our model using linear regression with sandwich robust standard errors. We estimated following DD model:

where y_cdp display the rate of the outcome of interest (per million persons) for country c, in day d, and 7-day period p. γ_c captures country-specific fixed effects that are time-invariant and δ_p captures period-specific fixed effects which are common across countries. Post is a dummy variable equal to 1 for Norway in post-lockdown periods and zero otherwise. α is the estimate of interest comparing changes in the outcome of interest in Norway before and after the lockdown with changes in Sweden over the same period. DD design relies on parallel trend assumption and additionally assume the constant effect for whole post-lockdown period. Plausibility of both these assumptions can be explored using event study design which was estimated using the following regression:

where 1(p-5=k) is an indicator for the 7-day periods relative to the lockdown period in Norway (p=5, March 26–April 1, 2020). We set 1(p-5=k) to zero for all observations in Sweden. We used the period prior to the lockdown (p=4) as the baseline. α_k is the estimate of interest comparing changes in the outcome of interest between period p (post-lockdown) and the baseline period in Norway with the same changes in Sweden. β_k is an estimate of pre-lockdown trends and if different from zero, then it suggests differential pre-lockdown trends meaning that the parallel trend assumption is unlikely to hold. Separate analyses for the RMDs and non-RMDs cohorts were implemented. We also conducted subgroup analyses by age groups (aged≥65 years and aged≥80 years) and in the osteoarthritis (OA) cohort. All analyses were conducted in Stata version 17.

A total of 744,326 (Skåne) and 3,076,824 (Norway) people aged≥35 years alive on 26 February 2021 were included. of these, 48.9% in Skåne and 59.5% in Norway had a RMD diagnosis. In addition, 28.4% and 26.2% of the RMDs cohort were diagnosed with OA in Skåne and Norway, respectively. Moreover, about 13% (11%) of the RMDs cohort and 6% (7%) of the non-RMDs cohort in Skåne (Norway) were aged≥80 years. In pre-lockdown period, there were, on average, 6.9 vs. 3.3 COVID-19 hospitalizations and 0.1 vs. 0.2 COVID-19 deaths per million persons per day in the RMDs cohort in Norway vs. Skåne (Figure 1). The corresponding rates in post-lockdown periods were 5.1 vs. 19.4 COVID-19 hospitalizations and 0.7 vs. 3.5 COVID-19 deaths per million persons per day in the RMDs cohort in Norway vs. Skåne.

The DD estimates (Table 1) showed that compared with Skåne, the strict lockdown in Norway was associated with 18.0 (95% CI 8.6, 27.3), 2.7 (1.5, 3.9), and 2.9 (–4.0, 9.7) fewer daily COVID-19 hospitalizations, COVID-19 death, and all-cause mortality per million persons, respectively, among people with RMDs. Corresponding reductions were 9.1 (95% CI 4.5, 13.7), 0.9 (0.4, 1.5), and 2.9 (–3.8, 9.7) among those without RMDs. The effects increased with age in both cohorts and the estimates for COVID-19 hospitalization were comparable for the OA cohort.

Event study revealed that the reductions in COVID-19 outcomes were evident from 1-2 weeks after the lockdown implementation and onwards (Figure 2). Moreover, for COVID-19 hospitalization, there were differences prior to the lockdown implementation suggesting potential reverse causality (i.e. Norway implemented the lockdown in response to growing COVID-19 hospitalization rate). Similar patterns were generally seen across subgroups (Figure 3 & Supplementary Figures A1-A2).

We found that the strict lockdown in Norway was generally associated with reductions in COVID-19 hospitalizations and (to a lesser extent) COVID-19 deaths. These effects were evident from 1-2 week after lockdown implementation and onward. However, we didn’t observe a conclusive impact on all-cause mortality rates. The lockdown impacts were greater for people with RMDs and older people.

The observed reductions in COVID-19 related outcomes in this study is consistent with previous studies comparing Sweden with other countries with strict lockdowns [3, 6], albeit with smaller effects is this study. For instance, Juranek et al. [3] reported 172 fewer COVID-19 hospitalizations per million persons during April 17–21, 2020 in Norway compared with Sweden versus 121 and 66 for the RMDs and non-RMDs cohorts, respectively, during the same period in our study. This smaller effect is possibly due to the fact that Skåne region was among the least affected regions in Sweden during the first wave of COVID-19. This might also explain no effect on all-cause mortality, even though Catalano et al. [2] also reported 963 excess non-COVID deaths in Sweden compared to Norway during 9 weeks following the lockdown but with a high uncertainty (p-value>0.05). It should be noted that differences in data sources, study population and design, and the definitions of pre- and post-lockdown periods and COVID-19 outcomes complicate cross-study comparisons. For instance, on contrary to our study, Catalano et al. [2] treated March 16-25 as post-lockdown and attributed the excess mortality during this period to the lockdown.

More pronounced effects for the RMDs than the non-RMDs cohort might reflect better adherence to the lockdown among persons with RMDs. It might also indicate higher adverse effects of COVID-19 among people with RMDs as recent studies reported [12-15] and hence greater gain from the lockdown. If so, this suggests that persons with RMD can be among risk groups prioritized for COVID-19 vaccination, even though further research is needed to identify those persons with RMDs with higher risk of COVID-19 adverse effects. Recently, England et al. [12] suggested vaccine prioritization for patients with rheumatoid arthritis receiving immunosuppressive therapies.

Several limitations should be acknowledged. We used register data to identify RMD status and the outcomes of interest which is prone to misclassification and coding errors. Due to the lack of data on specific cause of death in Skåne register, we classified a death as COIVD-19 death in those who died within 30 days from a COVID-19 hospitalization. This might underestimate COVID-19 death compared to the definition used by the Public Health Agency of Sweden, that is death within 30 days from a laboratory-confirmed COVID-19 test. The presence of pre-lockdown differences in COVID-19 hospitalization possibly reflects reverse causality [15] which hinder attributing the observed effects to the lockdown. In addition, people were taking voluntary social distancing measures regardless of the lockdown. Beside, Norway had higher level of testing and tracing COVID-19 cases compared with Sweden during the study period. These mean that the effects observed in the current study cannot be fully attributed to the lockdown especially if there were cross-country differences in voluntary actions taken by people. The low number of COVID-19 outcomes influenced the precision and uncertainty around our estimated especially in our subgroup analyses. Moreover, the low number of outcomes didn’t allow us to conduct further subgroup analysis.

Implementation of strict lockdown in response to the COVID-19 pandemic in Norway was associated with lower COVID-19 related hospitalizations and death, but no conclusive impact on all-cause mortality. Our findings confirm effect of strict lockdown in particular among those at higher risk of COVID-19 adverse outcomes such as older persons with RMDs.

CI: Confidence interval; DD: Difference-in-difference; ICD: International Statistical Classification of Diseases and Related Health Problems; OA: Osteoarthritis; RMDs: Rheumatic and musculoskeletal diseases

Acknowledgements Not applicable.

Author contributions: All authors contributed to study conception and design. AK performed the statistical analysis, and drafted the manuscript. AT, KM, FM, and ME participated in data acquisition. All authors contributed to the interpretation of the results, revising the manuscript critically for important intellectual content and approved the final version for submission.

Funding: This study was supported by the Foundation for Research in Rheumatology (FOREUM), the Swedish Research Council, the Österlund Foundation, and Governmental Funding of Clinical Research within National Health Service (ALF).

Data availability statement: The data supporting the findings of this study are available within the article and its Supplementary Information files or upon reasonable request.

Ethics approval and consent to participate: The study was approved by the Ethical Review Board, Sweden and the Regional Ethics Committee South-East, Norway. The need for individual informed consent was waived by the Ethics committees. The study was performed in accordance with the Declaration of Helsinki.

Consent for publication Not applicable.

Disclosure statement: ME and SBZ has received personal fees from Pfizer outside the submitted work. All other authors have no conflict of interest to declare.

Hale T, Angrist N, Goldszmidt R, Kira B, Petherick A, Phillips T, et al. A global panel database of pandemic policies (Oxford COVID-19 Government Response Tracker). Nat Hum Behav 2021;5:529-38.
Catalano R, Casey JA, Bruckner TA, Gemmill A. Non-COVID-19 deaths after social distancing in Norway. Eur J Epidemiol 2020;35:1021-4.
Juranek S, Zoutman FT. The Effect of Social Distancing Measures on Intensive Care Occupancy: Evidence on COVID-19 in Scandinavia. In: Norwegian School of Economics DoBaMS, ed. Bergen, Norway; 2020.
Conyon MJ, He L, Thomsen S. Lockdowns and COVID-19 deaths in Scandinavia. Covid Economics 2020;26:17-42.
Claeson M, Hanson S. The Swedish COVID-19 strategy revisited. Lancet 2021;397:1619.
Cho SW. Quantifying the impact of non-pharmaceutical interventions during the COVID-19 outbreak: The case of Sweden. The Econometrics Journal 2020;23:323-44.
Juranek S, Paetzold J, Winner H, et al. Labor market effects of COVID-19 in Sweden and its neighbors: Evidence from novel administrative data. Covid Economics 2020;42:143-63.
Wu Z, McGoogan JM. Characteristics of and Important Lessons From the Coronavirus Disease 2019 (COVID-19) Outbreak in China: Summary of a Report of 72314 Cases From the Chinese Center for Disease Control and Prevention. JAMA 2020;323:1239-42.
Team CC-R. Preliminary Estimates of the Prevalence of Selected Underlying Health Conditions Among Patients with Coronavirus Disease 2019 - United States, February 12-March 28, 2020. MMWR Morb Mortal Wkly Rep 2020;69:382-6.
Ziade N, El Khoury B, Zoghbi M, Merheb G, Karam GA, Mroue K, et al. Prevalence and pattern of comorbidities in chronic rheumatic and musculoskeletal diseases: the COMORD study. Sci Rep 2020;10:7683.
Swain S, Sarmanova A, Coupland C, Doherty M, Zhang W. Comorbidities in Osteoarthritis: A Systematic Review and Meta-Analysis of Observational Studies. Arthritis Care Res (Hoboken) 2020;72:991-1000.
England BR, Roul P, Yang Y, Kalil AC, Michaud K, Thiele GM, et al. Risk of COVID-19 in Rheumatoid Arthritis: A National Veterans Affairs Matched Cohort Study in At-Risk Individuals. Arthritis Rheumatol 2021;73:2179-88.
Williamson EJ, Walker AJ, Bhaskaran K, Bacon S, Bates C, Morton CE, et al. Factors associated with COVID-19-related death using OpenSAFELY. Nature 2020;584:430-6.
D'Silva KM, Jorge A, Cohen A, McCormick N, Zhang Y, Wallace ZS, et al. COVID-19 Outcomes in Patients With Systemic Autoimmune Rheumatic Diseases Compared to the General Population: A US Multicenter, Comparative Cohort Study. Arthritis Rheumatol 2021;73:914-20.
Kiadaliri A, Magnusson K, Turkiewicz A, Dell'Isola A, Runhaar J, Bierma-Zeinstra S, et al. Impact of the first wave of the COVID-19 pandemic on healthcare use in osteoarthritis: A population register-based study in Sweden. Osteoarthr Cartil Open 2022;4(2):100252.
Goodman-Bacon W, Marcus J. Using Difference-in-Differences to Identify Causal Effects of COVID-19 Policies. Survey Research Methods 2020;14:153-58.

Table 1. The impact of strict lockdown in Norway on daily rates of COVID-19 hospitalization/death and all-cause mortality: estimates from difference-in-difference analysis.

	RMD					Non-RMD
	Aged≥35 years	Aged≥65 years	Aged≥80 years	Osteoarthritis	Aged≥35 years		Aged≥65 years	Aged≥80 years
COVID-19 hospitalization	-18.0 (-27.3, -8.6)	-29.1 (-42.1, -16.2)	-46.1 (-64.5, -27.8)	-21.8 (-34.8, -8.9)	-9.1 (-13.7, -4.5)		-17.8 (-26.1, -9.6)	-26.9 (-41.8, -12.0)
COVID-19 death	-2.7 (-3.9, -1.5)	-5.8 (-8.3, -3.3)	-13.0 (-20.4, -5.6)	-4.6 (-6.9, -2.2)	-0.9 (-1.5, -0.4)		-3.7 (-5.7, -1.6)	-9.9 (-19.4, -0.4)
All-cause mortality	-2.9 (-9.7, 4.0)	-7.9 (-23.4, 7.7)	-21.5 (-59.7, 16.7)	-8.2 (-18.0, 1.5)	-2.9 (-9.7, 3.8)		-4.1 (-24.2, 16.0)	-29.7 (-95.7, 36.2)

RMD: Rheumatic and musculoskeletal diseases

Competing interest reported. ME and SBZ has received personal fees from Pfizer outside the submitted work. All other authors have no conflict of interest to declare.

Pandemic lockdown restrictions and COVID-19 hospitalization and deaths in patients with rheumatic and musculoskeletal diseases

Status:

Version 1

Abstract

Figures

Background

Methods

Results

Discussion

Conclusion

Abbreviations

Declarations

References

Tables

Additional Declarations

Supplementary Files

Status:

Version 1