COVID-19 Infection and Subsequent Psychiatric Morbidity, Sleep Problems & Fatigue: ... an English Primary Care Cohort, 2021, Chalder, Chew-Graham

[Sly Saint]

COVID-19 Infection and Subsequent Psychiatric Morbidity, Sleep Problems and Fatigue: Analysis of an English Primary Care Cohort of 226,521 Positive Patients

Abstract
Background: While some people who have had COVID-19 experience fatigue and sleep problems long after the acute phase of the illness, the proportion affected and the length of time symptoms persist remains uncertain. There are also concerns of an increase in psychiatric illness following COVID-19 infection; however evidence of a direct effect is inconclusive.

Methods: This was a retrospective cohort study using a UK primary care registry. Adults with a positive PCR COVID-19 test between 1st February and 8th December 2020 were matched to controls based on gender, general practice and year of birth. Separate matched cohorts were assembled for those with and without prior recorded mental illness, fatigue and sleep problems. Cox proportional hazard models estimated the association between a COVID-19 positive test and subsequent psychiatric morbidity (depression, anxiety, psychosis, or self-harm), sleep problems, fatigue or psychotropic prescribing, adjusted for comorbidities, ethnicity, smoking status and BMI. Interaction with age and area-level deprivation were tested for. Additional cohorts examined those with a negative COVID-19 test and those with influenza symptoms (and a negative COVID-19 test).

Results: After adjusting for observed confounders, there was an association between testing positive for COVID-19 (n=232,780) and almost all markers of psychiatric morbidity, psychotropic prescribing, fatigue and sleep problems. The adjusted hazard ratio for any indicator of incident psychiatric morbidity was 1.75 (95% CI: 1.56-1.96) and for new psychotropic prescribing 2.17 (95% CI: 2.00-2.35). However, there was an increased risk of incident psychiatric morbidity for those with a negative COVID-19 test of similar magnitude and a larger increase associated with having influenza.

Interpretation: Whilst COVID-19 infection appears to be causing sleep problems and fatigue, there is less clear evidence of an effect on subsequent psychiatric morbidity.

https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3858567

 

[rvallee]

In what world are fatigue and sleep problems psychiatric anyway? What a useless study. Also that's a lot of authors for such a pointless study. Clearly they have nothing useful to do and somehow I think that should be a concern.

2 notables: antidepressants and antianxiolytics are massively and needlessly overprescribed in this population and also this is the first paper featuring members of the BPS gang that does not feature "more research is needed" in its conclusions or interpretation that I can remember.

 

[dreampop]

Full article available as pdf. As expected, even with results that contradict their initial hypothesis, the authors manage to contort it all and bring it back to illness attribution and point the finger at health anxiety.

 

[Snow Leopard]

It turns out having a nasty infection also harms some people's mental health. Who would have guessed?

 

[Simon M]

I haven't finished reading it yet, but I really like this study, which seems to me to be done well. The hypothesis was that Covid would cause psychiatric morbidity. But they came up with a design to give that hypothesis a good shakedown.

Interpretation: Whilst COVID-19 infection appears to be causing sleep problems and fatigue, there is less clear evidence of an effect on subsequent psychiatric morbidity.

from the "research in context" box

while this study [indicates] increased risk of long-term psychiatric sequelae, evidence is provided indicating caution about the extent to which the increase is attributable to Covid 19 infection.

The key point was that there was also increased psychiatric morbidity (but not so much increased sleep problems and fatigue) in the control groups of people who tested negative. [Presumably, some of these people were positive but were not picked up by the test.]

The authors argue that other studies that found in effect may be picking up "unobserved confounding":

Some sources of confounding inevitably remain unobserved, or poorly measured, and in such cases the validity of course claims can be examined using a "negative control". Here, a variable with no conceivable effect on the outcome, but with a similar confounding structure, is substituted for the exposure and investigation. If the results in the negative control is similar to that observed using the primary exposure, then unobserved confounding is implicated.

A note on confounding
One possible example of confounding is occupation – people working in transport are more likely to be infected. Iff this was the case you could hypothetically wind up with a lot of bus drivers in the PCR positive cohort, and end up measuring things about bus drivers rather than the effects of Covid.

In what world are fatigue and sleep problems psychiatric anyway?

The authors clearly separate psychiatric morbidity from fatigue and sleep problems in their analysis and their write up. And they found no clear evidence for psych morbidity caused by covid, but good evidence for increased sleep problems and fatigue.

 

[Mithriel]

I admit that I will take a lot of persuading that these authors have changed their spots. It makes me look for ulterior motives.

while this study [indicates] increased risk of long-term psychiatric sequelae, evidence is provided indicating caution about the extent to which the increase is attributable to Covid 19 infection.

I can see this used to say that any increase in MUS is not because of infection with covid but because of the fear of getting infected and the anxieties of the pandemic. Anyone who has lived through lockdown now has a trauma to justify an FND diagnosis for instance.

What springs out is that people who had the genuine disease (and so have a genuine sick role) are mentally resilient but the weak willed lose themselves in anxiety and depression.

I would really like to be proved wrong.

 

[cassava7]

The authors note:

(...) however, we would caution against using convenience samples or internet-derived samples, where volunteers may be more mentally healthy than the general population, or more concerned about their health and more motivated about reporting it. (...) Given the highly specific circumstances of the pandemic and the vicissitudes of testing availability, the evidence to date on links between COVID-19 and subsequent ongoing symptoms of psychiatric morbidity, fatigue, and sleep problems – i.e. post-COVID syndrome or long COVID may contain important sources of bias. We conclude that the information to-date does not support a direct effect of COVID-19 and subsequent psychiatric morbidity and that more work is needed to understand if a causal connections exist and what services might anticipate in relation to future increases in clinical need.
​

From the OpenSAFELY data on the clinical coding of long Covid and the statistics on long Covid from the ONS, we know that long Covid is prevalent but heavily under-reported due to the lack of coding. Considering that the data source in this study was a primary care registry, it is conceivable that long Covid went undiagnosed/not coded for most patients with a positive Covid test, and was instead recorded as depression or anxiety, fatigue and sleep problems.

This would especially have held until recently, as public and medical awareness of long Covid was not as prominent earlier, and might contribute to explaining this finding: "However, there was an increased risk of incident psychiatric morbidity for those with a negative COVID-19 test of similar magnitude and a larger increase associated with having influenza."

Further, the increased prescription of psychotropic drugs in those who tested positive versus negative may also have originated from medical attitudes towards post-acute infectious symptoms (e.g. if considered to be psychosomatic). General practitioners might not have wanted to initiate a new treatment "for no reason", i.e. without proof of infection, and thus have conditioned prescribing new psychotropic drugs on a positive test.

On the other hand, if this explanation held, the hazard ratio might have been lower than 2.17 (and the CI wider than 2.00-2.35). Such an increase may also have been observed in the group with influenza and a negative Covid test (barring GPs minimizing the post-acute symptoms of flu vs. those of Covid and, in this case, not willing to initiate psychotropic treatment).

 

[Sean]

We conclude that the information to-date does not support a direct effect of COVID-19 and subsequent psychiatric morbidity and that more work is needed to understand if a causal connections exist...

Conceding that causality (and hence direction of causality) has not been established, instead of just assuming it has as per the normal BPS practice, is a step forward.

Albeit from a very long way behind, and very late in the game.

And way too late for Chalder, at least, to redeem herself.