Epi in the News

Hi Sujit,

Thank you very much for your valuable discussions.
I will try to answer the questions before reading the article.

1. What is the study design?
Observational Prospective Cohort Study.

2.For the 39% figure:

2.a. The population of interest: People who died from the selected individuals from the UK Biobank study, with an average age of 57 years, during the study period (12.6 years).

2.b. The Exposed Group: People who were never visited by friends or family, from the selected individuals from the UK Biobank study, with an average age of 57 years, during the study period (12.6 years).

2.c. The comparison group: People who were visited daily by friends or family from the selected individuals from the UK Biobank study, with an average age of 57 years, during the study period (12.6 years).

2.d. The outcome: Overall death or death from cardiovascular disease.
The outcome is an incidence measure, i.e. Incidence Risk.

3. RR is the Risk Ratio, according to the question the 39% is an increased risk the corresponding risk ratio for the figure 39% = 1 + 0.39 = 1.39, which means that there is increased risk of death of 1.39 times in the people who were not visited by friends or family (exposed group) in comparison to the people who were visited by friends or family (unexposed group).

4. People who were not visited by friends and family develop psychosocial burden. As a result, they may start to feel socially disengaged which may exert stresses that lead to some compulsive behaviours like excessive smoking and alcohol drinking, substances and drug abuse, lack of physical activities, and depression. All the before mentioned may eventually lead to death, with cardiovascular diseases being a major cause of death.

5. One potential confounder is age, as the researcher selected the group of an average age of 57 years, and there is increased risk of developing cardiovascular disease with the increase of age.

6. Although the RCT is ideal to confirm the causation, but we need to address the feasibility of running an RCT, taking into account the ethical consideration of the study, because the outcome of such a study might be harmful for the participants.

If the ethical committee approved an RCT in this context, then we may consider the following:

Population: Healthy individuals (approved after complete medical check), age between 25-35 years, with no history nor current smoking or alcohol drinking, with stable/ average level of income (to ensure they are from relatively same socioeconomic status), living in a specific catchment area (we can use for example a postal address).

Intervention: Deprive the intervention group (after random allocation) from any visit by friends and family for a specific period of time (e.g. 6 months)

Comparison: From the defined population some participants will be randomly allocated to the control/comparison group, and ensure regular visits by their friends and family during the study period (6 months)

Outcome: Cardiovascular disease related death or overall death over the study period (6 months).

Thanks,
Rajaa

1. What is the study design?

Cohort Study Design- Followed over time

2a. Who is the Population of interest?

 Adults across the UK from the UK biobank study with an ave. age of 57

2b. What group is 'Exposed'?
People who were never visited by friends or family

2c. What is the Comparison (unexposed) group?

Those who were visited daily by friends or family

2d. What is the Outcome? Is this a prevalence or incidence measure?

Incidence measure

3. For the 39% figure, what is the corresponding RR? What RR is this?
Risk is increased to 39% (100+39) = 139%
therefore, RR exposed is 139/100= 1.39 (an unadjusted RR)
1.39 means that individuals who were never visited by friends or family had a 39% higher risk of experiencing the outcome

4. What are the hypothesised mediating (causal pathway) mechanisms?

Casual pathways may include: Stress, anxiety, and depression, which can lead to alcohol intake, drug use, overeating, and a sedentary lifestyle. These behaviors negatively impact cardiovascular health and can cause death.

5. Without reading the journal article, can you propose and justify at least one confounder?

Underlying health status could be a confounding factor.

6. RCT might be ethically wrong here. Having to assign individuals to receive or not receive visits randomly might lead to isolation, distress, and some other negative outcomes.
Population: would be adults with no visits and social connections
Comparison: Adults who receive regular visits
Outcome: Cardiovascular disease death

Good afternoon to Sujit and all!
What an interesting and important topic

1. What is the study design? it is a descriptive, analytical (because they compare different expossures any visit-x visits) and longitudinal prospective, that would match with a cohort study, as my colleagues have suggested.

People who were never visited by friends or family were 53% more likely to die from cardiovascular disease and had a 39% increased risk of death compared with those who were visited daily.

2. For the 39% figure, let's try something like a PICO, but more like a PECO:

2a. Who is the Population of interest? People around 57 years old
2b. What group is 'Exposed'? The exposed group would be the ones with the risk factor which in this case it is supossed to be loneliness (no visits).
2c. What is the Comparison (unexposed) group? the viceversa of the 2b.
2d. What is the Outcome? Death.
Is this a prevalence or incidence measure? They are using incidence risk as death rate. The key is in the denominator because is number of deaths in the numerator by those at risk of the exposure. Although we talk about death rate it is not a real rate as we are not saying years-person

3. For the 39% figure, what is the corresponding RR? 1.39 What RR is this? it is the incidence risk of dying of all cause being lonely compared to those not living alone

4. What are the hypothesised mediating (causal pathway) mechanisms? they don´t study casualty, but they have hypothesised that "“It could be that people who are more socially isolated may have some more unhealthy behaviours like smoking or high alcohol intake, for example.” He also suggested that not having someone to help take them to the doctor or encouraging them to seek help when needed, as well as direct biological effects on the immune system, could be factors."

5. At least from just reading this new article, I feel like the observed association could be partly or wholly explained by confounding factors. Without reading the journal article, can you propose and justify at least one confounder? (Feel free to check the article afterwards...) I don´t know if they took into account any other variables, but I would say as demographics, socioeconomic status is very important, educational level as well. then diabetes, hypertension, smoke, alcohol intake, type of diet, anxiety, depression... there are many things to consider in these case

6. This was an observational study. To confirm causation, do we need to run an RCT? What is the PICO (Population, Intevention, Comparison, Outcome) for your RCT? I wouldn´t run an RCT because I find it very unethical, loneliness it is not only a matter of health but a matter of comunity and humanity, so I would never make people experience loneliness on purpose. Taking all of that into account I would do the following.
P: people between 45 and older
I: People receiving daily visitis
C: normal situation before the trial
O: hospitalization, death, and other dependent variables as quality of life, levels of anxiety and depression (it would be cruel to let them die if I find a positive effect of daily visits)

1. What is the study design?
The study design is a cohort study because they were followed over time (12.6 years)

2. For the 39% figure, let's try something like a PICO, but more like a PECO:
2a. Who is the Population of interest?
People across the UK from five different kinds of social connection with an average age of 57.
2b. What group is 'Exposed'?
People who were never visited by friends or family
2c. What is the Comparison (unexposed) group?
People who were visited daily.
2d. What is the Outcome? Is this a prevalence or incidence measure?
Risk of death, Incidence measure
3. For the 39% figure, what is the corresponding RR? What RR is this?
The corresponding RR is 100+39 = 139% = 1.39
Therefore, people whose friends or family visited died 1.39 times more likely than those who were visited daily.
4. What are the hypothesised mediating (causal pathway) mechanisms?
The possible causal pathway from this context is stress or depression from staying alone, which gives rise to these people who were not visited being heavy alcohol and heavy smokers because they want to relieve their stress.

5. Can you propose and justify at least one confounder? (Feel free to check the article afterwards)
People probably have some abnormal mental health history.
The researcher should regard this condition because mental status could affect people's behaviour and body health. To justify, one confounder is to calculate RR among people by mental health history (each stratum). If the results differ from crude RR but are similar to each other, this is evidence of confounding.

To confirm causation, do we need to run an RCT? What is the PICO (Population, Intervention, Comparison, Outcome) for your RCT?
In my opinion, this study cannot conduct the RCT because of an ethical issue; however, if we need to run an RCT.
Population : Adult in the UK
Intervention : People are not visited by anyone.
Compararison : People are visited by friends or family.
Outcome : Cardiovascular disease instead of death

Hello Sujit,

Here are my responses:


1. What is the study design? – Prospective cohort study (subjects were followed for 12.5 yrs)

People who were never visited by friends or family were 53% more likely to die from cardiovascular disease and had a 39% increased risk of death compared with those who were visited daily.

2. For the 39% figure, let's try something like a PICO, but more like a PECO:
2a. Who is the Population of interest? – Adults across the UK.
2b. What group is 'Exposed'? – Exposed are individuals not experiencing any form of social interaction
2c. What is the Comparison (unexposed) group? - Individuals living with someone who saw friends or family daily.
2d. What is the Outcome? Is this a prevalence or incidence measure? - The authors measured incidence in this prospective cohort study. The outcome of interest was death.

3. For the 39% figure, what is the corresponding RR? What RR is this? (was not sure about this, how we get %)
This is a risk ratio (the ratio of incidence of death among visited daily devided bythe incidence of death among never visited). 39% is an increased risk of dying among those who never visited compared with those who had daily visits.

4. What are the hypothesised mediating (causal pathway) mechanisms? – These could be presence of unhealthy behaviours like smoking or high alcohol intake

5. At least from just reading this new article, I feel like the observed association could be partly or wholly explained by confounding factors. Without reading the journal article, can you propose and justify at least one confounder? (Feel free to check the article afterwards...) –

Consumption of high alcohol intake should be examined among exposed subjects that could serve as underlying factor for the risk of death. Regular intake of alcohol worsens health condition of adults particularly when there is a presence of comorbidities (diabetes, HIV, TB etc.) or other factors (education level, income etc.). Stratification (if possible) in this case will be a relevant approach to avoid overestimated results between the exposure and outcomes of interests.

6. This was an observational study. To confirm causation, do we need to run an RCT? What is the PICO (Population, Intevention, Comparison, Outcome) for your RCT?

I do agree with unethical aspect, however, if conducting RCT will be necessary, then we could compare the findings with the results from Biobank study. If this is the objective, then the target population of interest would include adults of the same age (57). Individuals will be randomly assigned to two groups: 1)those who will receive any form of social interaction with family or friends and 2)comparison group who will be isolated from any form of social interaction. Outcome of interest will defined as the risk of death

Hi Sujit
Couldnt this be an incidence rate with the denominator as person-years?. I have not read the whole article, but there might be individuals who were not visited by anyone for a given time period, but were at another time period

Hi Sujit,

The following are my answers:

1. This is a prospective cohort study.

2. 2a. The population of interest is adult people across the UK (average age 57 years and included in the UK Biobank study).

2b. The exposed group is adult people who were never visited by friends or family.

2c. The comparison (unexposed) group is adult people who were visited daily by friends or family.

2d. For the 39% figure, the outcome is all-cause mortality.
This is an incidence measure.

3. For the 39% figure, the corresponding RR is 1.39.
This RR is rate ratio for mortality between exposed and unexposed where the rates have been calculated as deaths per 100 person-years by dividing the number of deaths within each group (exposed/unexposed) by the total number of years of follow-up (12.6 years) in each group and multiplied by 100.

4. The hypothesised mediating (causal pathway) mechanisms are:
(i) socially isolated and lonely people are more likely to adopt unhealthy behaviours like smoking or high alcohol intake, and suffers from related health problems; (ii) social isolation and loneliness leads to inadequate health seeking behaviour due to not having someone to help them or encouraging them to seek help when necessary; and (iii) it also has direct biological effects on the immune system. All these can cause increased mortality. 

5. Age is a confounding factor here. The average age of population included in this study was 57. Age is associated with social isolation and loneliness. Age is also an independent risk factor for cardiovascular diseases, diabetes, depression, dementia, etc., and related mortality. Age is not in the causal pathway between social isolation or loneliness and mortality.

6. We should not run an RCT here to confirm causation, as it will involve ethical issues. It is not ethical to randomly assign people to an intervention group (never receiving visits by friends or family) and the comparison group (receiving daily visits by friends or family).

PICO for the RCT:
P (Population): Adult population aged 50 or over
I (Intervention): Never receiving visits by family or friends
C (Comparison): Receiving daily visits by family or friends
O (Outcome): Cardio-vascular mortality and all-cause mortality.


Thanks!
Madhutandra