An Epidemiology Study On The Effect Of Housing Improvements On Smoking: Findings, Bias, And Causality Criteria
Quasi-experimental and longitudinal study design
In this epidemiology study conducted by Bond et al. (2013), a ‘Quasi- experimental’, ‘longitudinal’ study design was applied for this investigation.
One of the key features for Quasi- experimental study is the lack of random assignment to either control or treatment. Another key feature is the presence of ‘pre and post testing’. In this experiment, both of the characteristics were fulfilled by the investigators (Wells & Yang, 2008). The subjects for this investigation were not randomly assigned and the subjects were pre tested before the intervention and post tested after the intervention.
Longitudinal study is a kind of study where data are collected during a period of time and in this investigation data were collected over a period of two year.
In this research investigation two research questions were presented and based on this research questions following two hypotheses can be stated:
- HI (Housing Improvements) will lead to an intention to quit smoking or reduction in smoking.
- An intention to quit smoking or reduction in smoking will happen as a direct consequences of reduction in stress and improvement of mental health due to HI (Housing Improvements).
The study factors for this study are the housing improvements like doors, external cladding, roof improvement, kitchens, bathrooms, and heating.
Surveys were used to collect the data for the study factors. The groups were measured and divided on the basis of housing improvements. The participants were asked whether they had experienced any housing improvements mentioned in the above paragraphs in the past two years. Those who reported to experienced housing improvements were placed in the HI group and those who did not reported any experiences of housing improvements were placed in the non-HI group.
The outcome of this investigation is whether there is an intention to quit smoking or reduction in smoking. The factors which might be affecting this outcome are anxiety, stress, depression, as well as socio- demographic variable like ethnicity, gender, education, and age.
Anxiety, stress, and depression can be used to determine the status of mental health and mental assessment tool were used to measure this aspects. SF- 12v2 assessment tool was used to assess the status of mental health. The composite score for mental health assessment was determined using 12 questions as per as the SF- 12v2 procedure. Time period for these questions was ‘over the past four weeks. Economic status of the participants was determined on the basis of participant’s economic activities.
The main findings of the Table 2 was whether participants from HI group were intended to quit smoking in comparison with the participants from the non- HI group. The findings from the Table 2 suggest that the participants from HI group were twice as likely quit smoking when compared with the participants from non- HI group. The p- values of the findings were statistically significant as well. Therefore, it can be deduced that the housing improvements has a significant on the intention of quit smoking and hence, it can be said that the study factors are associated with outcome factors. These findings remained significant even after the addition of socio- demographic parameters.
Characteristics of Quasi-experimental design
The authors of this investigation determined the participants of this investigation through random sampling to avoid the selection bias. The participants were randomly selected from post office addresses. From the selected addresses, one adult (age greater than 16) person was randomly chosen with their consent. However, selection bias could be introduced in this study design (Allcott, 2015). The authors have selected 14 disadvantaged localities in Glasgow. This was determined using Scottish Index of Multiple Deprivation. From this, it can be seen that the authors have already pre- defined the criteria’s for locality selection. However, a society does not only contain the deprived population. Therefore, there might be chance of unequal distribution of sample selection. However, people from outside the chosen locality also have the habit of smoking. Additionally, people from higher economic status have more chance to experience housing improvements. Outcome factors like stress, depression also presents in these communities. Therefore, the far more unbiased results could be obtained if all other communities were involved in the study design.
In this study, information bias might play a role in the concluding results. The authors have collected data in pre- test and post- test method. There was no mention of the time frame of the occurrence of housing improvements. They have not set up any criteria in this regard. For example, this was longitudinal study over a period of two years. A participant might experience housing intervention one month after the start of the investigation, six month after, one year after or just before the completion of the two year and it is quite obvious that the each participant will experience this intervention at different time frame. Therefore, all the participants will not experience the effect of intervention for the same duration of time and effect of housing intervention on smoking, if any, is bound to be erroneous. Hence, from the above discussion, it can be said that the information bias might have an effect on the final data.
Confounders are variables which influences both the dependent and independent variables causing an unauthentic interpretation of the data (Raghunathan, Miller & Rashid, 2015). Confounders are present in this study as well and the authors have identified those confounders and take measures to nullify their effects. The confounders which have been identified in this study are socio- demographic parameters like ethnicity, age, gender as well as their employment status (employed or retired) and educational status (attained school or not). The reason behind the consideration of the confounders by the authors is to reduce the effect of unauthentic interpretation or erroneous results. In the data presented in Table 2, the authors have reported that the relation between the study factor and outcome factor remained significant even after the inclusion of confounders. The relation between the study factor and outcome factor was that the housing improvements enhance the chance of intention to quit smoking. Therefore, in this scenario, the confounders had no effect on the relationship between study factors and outcome factors. The possible explanation behind this is that the study factor is the sole contributing factor to the outcome of this investigation’s findings.
Subjects and Housing Improvements
The second research question which was posed by the author was “could this be explained by improvements in mental health and/ or reductions in stress subsequent to the HI?” In the presented data in Table 3 the authors have answered this research question. In order to do that, the authors have constructed 4 multivariate models for examination. They have reported that the odds ratio for HI remains significantly high for all the models. Consultation with GP regarding emotional problem strongly associated with the intention to quit smoking, however, it is not significantly associated with the housing improvements. In their third assessment model, they have shown that mental health managed to improve a person’s symptoms significantly, although, worsening of symptoms is only marginally significant. Similar like previous scenario, intention to quit smoking is not associated with housing improvements. In a nutshell, mental health is positively associated with the experience of housing improvements but negatively associated when considering about the intention to quit smoking. Improvements in symptoms like anxiety or depression is closely associated with intention to quit smoking but not with housing improvements. The author’s research question was to determine whether mental health could be associated with housing improvements and they have explored this area in detail.
In this scenario, I agree with the author that the high prevalence of smoking rate in the group is more likely due to the chance factor and it is most likely not related to the factor which is the receipt of housing improvements. The reason behind is that the smoker does not necessarily seek housing improvements as well as nor the housing improvements were provided based on the smoking characteristics. Therefore, the reason behind the higher smoking rate than national smoking rate in these two groups happened due to chance factor and nothing more. Although, studies have shown complete opposite findings where there is a strong correlation between reduction in smoking prevalence and housing improvements (Blackman et al., 2001). However, the study design of these two studies is different and this might be the reason behind the difference in findings.
Bradford Hill has set out nine different criteria for the evaluation of causality in epidemiological study (McDonald & Strang, 2016). In this study, three of these criteria can be established. These criteria are Consistency, Temporality, and Coherence. Consistency can be established in this regard as the authors have demonstrated the fact that there is a consistent observational finding between the reduction in smoking or intention to quit smoking and housing improvements. The criteria Temporality can be established only if the effect occurs after the cause. This is also is in line with this study’s findings. The outcome, intention to quit smoking or reduction in smoking, happened after the cause which is housing improvements. Similar like previous two criteria, coherence can also be established in the context of this study. This study has reported the relationship between intention to quit smoking with housing improvements and studies investigating in this area also presented similar kind of findings (Helms, King & Ashley, 2017).
The participants of this study were selected from the community which are below lowest 15 per cent economical deprivation cut off according to Scottish Index of Multiple Deprivation. All of the participants were from fourteen neighborhoods belongs in the fore mentioned criteria. No participants were chosen from any other community. Henceforth, the findings of this study can be generalized to the individuals who are belong to the lowest 15 per cent economical deprivation cut off according to Scottish Index of Multiple Deprivation.
References:
Allcott, H. (2015). Site selection bias in program evaluation. The Quarterly Journal of Economics, 130(3), 1117-1165.
Blackman, T., Harvey, J., Lawrence, M., & Simon, A. (2001). Neighbourhood renewal and health: evidence from a local case study. Health & place, 7(2), 93-103.
Bond, L., Egan, M., Kearns, A., Clark, J., &Tannahill, C. (2013). Smoking and intention to quit in deprived areas of Glasgow: is it related to housing improvements and neighbourhood regeneration because of improved mental health?. J Epidemiol Community Health, 67(4), 299-304.
Helms, V. E., King, B. A., & Ashley, P. J. (2017). Cigarette smoking and adverse health outcomes among adults receiving federal housing assistance. Preventive medicine, 99, 171-177.
McDonald, R., & Strang, J. (2016). Are take?home naloxone programmes effective? Systematic review utilizing application of the Bradford Hill criteria. Addiction, 111(7), 1177-1187.
Raghunathan, K., Miller, T. E., & Rashid, A. M. (2015). Confounders versus mediators: an important distinction. Anesthesiology: The Journal of the American Society of Anesthesiologists, 123(1), 234-234.
Wells, N. M., & Yang, Y. (2008). Neighborhood design and walking: a quasi-experimental longitudinal study. American journal of preventive medicine, 34(4), 313-319.