Potential gains in life expectancy from reducing amenable mortality among people diagnosed with serious mental illness in the United Kingdom.

BACKGROUND: To estimate the potential gain in life expectancy from addressing modifiable risk factors for all-cause mortality (excluding suicide and deaths from accidents or violence) across specific serious mental illness (SMI) subgroups, namely schizophrenia, schizoaffective disorders, and bipolar disorders in a Western population.
METHODS: We have used relative risks from recent meta-analyses to estimate the population attribution fraction (PAF) due to specific modifiable risk factors known to be associated with all-cause mortality within SMI. The potential gain in life expectancy at birth, age 50 and age 65 years were assessed by estimating the combined effect of modifiable risk factors from different contextual levels (behavioural, healthcare, social) and accounting for the effectiveness of existing interventions tackling these factors. Projections for annual gain in life expectancy at birth during a two-decade was estimated using the Annual Percentage Change (APC) formula. The predicted estimates were based on mortality rates for year 2014-2015.
RESULTS: Based on the effectiveness of existing interventions targeting these modifiable risk factors, we estimated potential gain in life expectancy at birth of four (bipolar disorders), six (schizoaffective disorders), or seven years (schizophrenia). The gain in life expectancy at age 50 years was three (bipolar disorders) or five (schizophrenia and schizoaffective disorders) years. The projected gain in life expectancy at age 65 years was three (bipolar disorders) or four (schizophrenia and schizoaffective disorders) years.
CONCLUSIONS: The implementation of existing interventions targeting modifiable risk factors could narrow the current mortality gap between the general and the SMI populations by 24% (men) to 28% (women). These projections represent ideal circumstances and without the limitation of overestimation which often comes with PAFs.

Introduction

While life expectancy of the general population has increased steadily over the past century, people with serious mental illness (SMI), including diagnoses of schizophrenia, schizoaffective disorder, or bipolar disorder, have a reduced life expectancy of 13 years for men and 12 years for women.[1] Around a fifth (20%) of the excess mortality in people with SMI is due to non-natural deaths such as suicide and accidents but most deaths can be ascribed to common diseases such as cardiovascular disease (CVD), respiratory illnesses, type 2 diabetes (T2DM), cancer, and digestive disorders.[2-4] These diseases, in their turn, are partly attributable to modifiable unhealthy lifestyle factors (e.g. smoking, sedentarism, obesity, dyslipidemia), adverse social context (e.g. social isolation, social deprivation), and suboptimal healthcare use and efficiency (e.g. prevention, treatment adherence).[5-11] These modifiable risk factors therefore provide opportunities for interventions to reduce the excess premature mortality experienced by people with SMI.[7, 12, 13] Currently, there is limited evidence about the potential gain in life expectancy among people with SMI by addressing multiple modifiable risk factors for all-cause mortality. In this study we have quantified the potential gain in life expectancy at birth and adult (ages 50 and 65) years that might realistically be achieved by evaluating the population attributional fraction due to specific modifiable risk factors and adjusting these estimates by the published effectiveness of interventions from international studies to address these factors. Projection of the expected annual increase in life expectancy at birth over future decades can help monitor progress towards meeting set goals across specific SMI groups. Therefore, this paper employed the life table method to estimate potentially attainable gain in life expectancy over the life course for specific SMI subgroups and projected the gains over a two-decade period.

Materials and methods

Data source

The existing literature was searched to estimate the prevalence of modifiable risk factors for natural causes of all-cause mortality within the SMI population, the relative risk (RR) for all-cause mortality attributed to those risk factors, and the effectiveness of health interventions targeting modifiable risk factors within the SMI population. SMI was broadly defined to include schizophrenia, schizoaffective disorders, and bipolar disorders. We searched PubMed and reviews-focused electronic databases (e.g. Cochrane Database of Systematic Reviews, DARE, DoPHER) to identify the most recent meta-analyses with relevant quantitative data between 2008 and 2018. Where no meta-analyses were identified, data from the most recent observational studies with representative populations were used. To facilitate comparison, priority was given to meta-analyses and observational studies with comparative data available across all three SMI subgroups. Where schizophrenia and schizoaffective disorders were grouped together or when only overall estimates were provided (rather than SMI subgroup specific estimates), we have assumed the same estimate within each SMI subgroup. A description of the resources used to derive the study estimates is provided in S1 Table in S1 File.

Serious mental illness

There is inconsistent definition of SMI in practice, [14-16] however, the term commonly refers to people diagnosed with schizophrenia, schizoaffective disorders, and bipolar disorder.[1, 2, 17] Major depressive disorders are also considered SMI, but not consistently, and here we have decided to use a narrower definition and consider SMIs with superior evidence about drivers of excess mortality.[18] While these disorders share some common characteristics (i.e. overlap in clinical and genetic attributes), core risk factors and associations, they also possess distinctive clinical features and healthcare burden.[19-24] For instance, depression and panic disorders were suggested to be more severe within bipolar disorder, while lack of insight and early age onset were more common within schizophrenia.[20, 23] Since these distinctive clinical features may lead to differences in comorbidity rates, the three SMI disorders may also present with distinct mortality associations. We therefore considered schizophrenia, schizoaffective disorders, and bipolar disorders as separate entities for the present study, though we acknowledge the substantial overlap in the probable drivers of excess mortality.[25]

Modifiable risk factors

The present study focused on natural causes of death and was restricted to risk factors known to be amenable and preventable within the SMI population. In the context of the present study, a modifiable risk factor was defined as an event known to directly or indirectly increase the probability of death and for which there are proven methods to reduce the impact on all-cause mortality in people with SMI.[26] These factors include behavioural risk factors (e.g. smoking, physical inactivity, obesity, poor diet, substance abuse) and healthcare-related factors (e.g. uptake of public health interventions, uptake of effective treatment, access to healthcare resources), as well as upstream socioeconomic determinants (e.g. social exclusion, stigma, deprivation). A multicontextual view of avoidable mortality was favoured since it spans the different levels of explanation for the reduced life expectancy within the SMI population as defined here. It also facilitates the identification of proven interventions to reduce the mortality gap between people with and without SMI.[27]

Statistical analysis

Prevalence of modifiable risk factors, relative risk for all-cause mortality, and the effectiveness of existing interventions were identified through systematic searches of published data. To determine the proportion of mortality burden attributable to each modifiable risk factor we computed the population-attributable fraction (PAF) from the prevalence (Pcs) of modifiable risk factors and unadjusted relative risk (RRcs) of all-cause mortality associated with each risk factor separately, using the formula:

The PAFs for all-cause mortality associated with a modifiable risk factor were computed using the prevalence and RR estimates obtained from meta-analysis or individual studies. The PAF is interpreted as the proportion of mortality risk that could be eliminated from the SMI population, if the exposure to the risk factor was eliminated or reduced. To allow for the overlap between different modifiable risk factors in influencing all-cause mortality, we have also estimated the combined impact of multiple modifiable risk factors within each group of determinants (e.g. behavioural, healthcare, social) and across all modifiable risk factors using, [28] where r is the risk factor index. While the formula assumes no interaction effects, it ensures that the PAF for the combined contribution of modifiable risk factors to all-cause mortality is not greater than 1.[28]

To estimate likely gain in life expectancy from specific interventions in SMI, we adjusted the PAF value of modifiable risk factors by the effectiveness of interventions aimed at reducing the rate of modifiable risk factors within the SMI population. For example, if the PAF for smoking was 46% (based on (1) above), then this figure was weighted by the effectiveness of smoking cessation interventions in people with SMI. If the effectiveness of smoking cessation among people with SMI would be around 36%,[29] then the likely gain in life expectancy at birth with respect to smoking would be based on the assumption that 17% (0.46 weighted by 0.36) of smoking-related mortality risk could be potentially eliminated from people with schizophrenia. Data on the effectiveness of interventions were only available at aggregate level.

Abridged period life table, based on the Chiang’s method,[30] for 2014–2015 were used to describe the base case population for life expectancy and all-cause mortality rates in United Kingdom (UK) adults with SMI. The 2014–2105 period represents the most recent data (2018) released by the National Health Service Outcome Framework Indicators (NHS QoF),[31] which includes aggregated mortality rate and population data into 5-year age intervals for SMI adults 18 to 74 years of age. For SMI adults aged 75 years or older, the mortality rates were calculated based on 2014–2015 mortality data among adults of similar age from the general population. Specifically, we have multiplied the mortality rate for 2014–2015 among adults aged 75 years or over from the general population by the increased mortality risk among adults aged 75 years or over with SMI (e.g. 2.5 for 75–79, 1.9 for 80 to 84, and 1.3 for 85+ year-group) using published data.[32] For instance, the mortality rate in 2014–2105 for SMI adults aged 75 to 79 years was estimated to be 2.5 times greater (0.0835) than the 0.0334 mortality rate experienced by 75–79 years old adults from the general population during 2014–2015. Because younger people (<15 years of age) are unlikely to receive a diagnosis of SMI,[1] we have used the UK all-cause mortality rates for the under 18 year age group in substitution. Separate abridged life tables were constructed for each modifiable risk factor. These mortality rates were used to create a cause-eliminating life table to estimate the potential gain in life expectancy at birth if modifiable risk factors could be reduced or eliminated within specific SMI subgroups.[33] The gain in life expectancy represents the change in life expectancy at birth that would be obtained under the hypothetical reduction of the cause (modifiable risk factors) of premature mortality within specific SMIs subgroups. Let 65.9 years denote the life expectancy at birth (age 0) from actual life table for a person with schizophrenia, and 68.3 year denote life expectancy at birth from smoking(cause)-eliminated life table for the same person. Then the gain in life expectancy at birth due to smoking(cause) elimination or reduction is 2.4 years (68.3–65.9). This approach has been widely employed to estimate the potential gain in life expectancy if a specific cause could be reduced or eliminated.[34] To capture potential gain in life expectancy across the life course and to allow for the fact that people are diagnosed with SMI at different life stages,[35] we have estimated two additional scenarios: (I) potential gain in life expectancy among people with SMI who survive to age 50, and (II) potential gain in life expectancy among people with SMI who survive to age 65. For (I), the analyses estimated the average remaining years of life that SMI survivors to age 50 were expected to live if mortality levels at each age-group over 50 remained constant.[36] A similar approach was used for age 65 (II).[37]

To quantify the expected average annual amount of change in life expectancy, we have divided the total gain in life expectancy at birth by the number of years during a specific period. Given the stable mortality rates and modest change in life expectancy over the past decade within the general population,[38] we have considered a 20-year period as a reasonable timeframe for projecting likely average annual amount of change in life expectancy among people with SMI. In addition, we computed the arithmetic average percentage of change in life expectancy during a 20-year period by dividing the average amount of annual change by the expectancy value at the beginning of the period, using the following formula: [33]

Lex2030/Lex2018 = ert(20)                (3)

transform (1) in ln: ln(Lex2038 / Lex2018) = rt

dividing (2) by (1): ln(Lex2038 / Lex2018)/t = r

Where, Lex2038, represent life expectancy in 2038, and Lex2018 refers to life expectancy in 2018., r represents the Annual Percentage Change (APC), and t the number of years. In sensitivity analyses, we applied average all-cause mortality rates over a five-year period (2009/2010 to 2014/2015) using mid-interval (2011/2012) population as the denominator to assess the robustness of our findings. We used Stata vers. 15 and Excel to analyse the data.

Results

Table 1 shows the estimates for the prevalence of modifiable risk factors within specific SMI subgroups. According to most recent meta-analyses,[39, 40] around 59% of people with schizophrenia and 49% of those with bipolar disorder are current smokers. Accelerometry-based evidence indicated that sedentary behaviour was common among 81% of people with schizophrenia and 78% among those with bipolar disorders[41, 42]. The uptake of public health interventions (i.e. screening, preventative initiatives) was also poor, with around 50% of people with SMI not being screened for CVD risk factors.[39] Low rates of receiving specialist treatment following a major life event were identified, with 47% of the SMI population not receiving revascularisation procedures following a myocardial infarction, for instance.[43] Regarding social context, around 65% and 55% of people with schizophrenia or bipolar disorders, respectively, reported the experience of stigma.[44, 45]

Table 1

Prevalence of modifiable risk factors within specific SMIs and the general population.

	Schizophrenia		Bipolar disorder		Schizoaffective disorders
	%	RR	%	RR	%	RR
LIFESTYLE FACTORS
Smoking	59	2.45	49	1.57	59	2.50
Sedentary	81	1.66	78	1.66	81	1.66
Diet	69	1.53	69	1.53	69	1.53
Obesity	50	1.47	50	1.47	50	1.47
Substance abuse	47	1.78	45	1.89	54	1.78
Metabolic syndrome	33	2.10	32	1.70	35	2.10
HEALTHCARE FACTORS
Clozapine/Lithium	24	1.88	29	1.70	24	1.88
Healthcare access	46	1.34	46	1.11	46	1.34
Treatment disparity (e.g. revascularisation)	47	1.15	47	1.15	47	1.15
Screening uptake	50	1.18	50	1.18	50	1.18
SOCIAL FACTORS
Social deprivation	30	1.90	26	1.30	26	1.90
Social isolation	32	1.19	21	1.19	32	1.19
Stigma experience	65	1.12	55	1.12	49	1.12

RR–relative risk of mortality given the risk factor

Lifestyle risk factors

Table 2 shows potential gains in life expectancy at birth, and ages 50 and 65 years associated with tackling modifiable risk factors for all-cause mortality in specific SMIs. Smoking-related conditions accounted for 46% of all deaths in schizophrenia and 22% of all deaths in bipolar disorders (based on formula (1) above). Based on age-aggregated effectiveness of existing smoking-cessation interventions, smoking-targeted initiatives have the potential to extend the life expectancy at birth by on average two years and five months within schizophrenia or schizoaffective disorders, and by one year and one month within bipolar disorders. Potential gain in life expectancy at age 50 related to smoking-targeted interventions, was two years within schizophrenia or schizoaffective disorders, and nine months among bipolar disorder. Likewise, potential gain in life expectancy at age 65 from smoking-targeted interventions was one year and seven months within schizophrenia or schizoaffective disorders, and seven months among bipolar disorder. Sedentary behaviour was the most potent determinant of avoidable mortality within bipolar disorders, accounting for just over a third (34%) all-cause mortality (using formula (1) above). Physical inactivity-focused interventions could extend life expectancy at birth among the SMI subgroups by on average one year and three months. Under an additive scenario, tackling behavioural and metabolic determinants of avoidable mortality would lead to a potential gain in life expectancy at birth of four (bipolar disorders) or six (schizophrenia) years. The combined gain in life expectancy at age 50 years was three (bipolar disorders) or five (schizophrenia) years, while at age 65 years it was two (bipolar disorders) or four (schizophrenia) years.

Table 2

Potential gains in life expectancy at birth and ages 50 and 65 years associated with modifiable risk factors for all-cause mortality in specific SMIs.

	Schizophrenia					Bipolar disorders				Schizoaffective disorders
	ES (%)	PAF (%)	LYG–age Birth 50 65			PAF (%)	LYG- age Birth 50 65			PAF (%)	LYG- age Birth 50 65
LIFESTYLE MODEL
Smoking	36	46	2.5	2.1	1.7	22	1.1	0.9	0.7	46	2.5	2.1	1.7
Sedentary	25	35	1.3	1.0	0.8	34	1.3	1.0	0.8	35	1.3	1.0	0.8
Diet	31	27	1.1	0.9	0.7	27	1.1	0.9	0.7	27	1.1	0.9	0.7
Obesity	24	19	0.7	0.6	0.5	19	0.7	0.6	0.5	19	0.7	0.6	0.5
Substance abuse	17	27	0.7	0.6	0.5	29	0.7	0.6	0.5	30	0.7	0.6	0.5
COMBINEDb		85	5.4	4.4	3.6	78	3.8	3.1	2.5	85	5.4	4.4	3.6
Metabolic syndrome	30	27	1.1	0.9	0.7	18	0.8	0.6	0.4	28	1.1	0.9	0.7
COMBINED (lifestyle)		89	5.8	4.7	3.8	82	3.8	3.0	2.4	89	5.8	4.7	3.8
HEALTHCARE MODEL
Clozapine/Lithiumb	13/38	17	0.3	0.2	0.2	17	0.9	0.7	0.6	17	0.3	0.2	0.2
Healthcare access	38	14	0.7	0.6	0.5	5	0.3	0.2	0.2	14	0.7	0.6	0.5
Treatment disparity	37	7	0.4	0.3	0.3	7	0.4	0.3	0.3	7	0.4	0.3	0.3
Screening uptake	41	8	0.4	0.3	0.3	8	0.4	0.3	0.3	8	0.4	0.3	0.3
COMBINED		37	0.7	0.5	0.5	36	0.8	0.6	0·5	37	0.7	0.5	0.5
SOCIAL MODEL
Social deprivation	23	21	0.7	0.6	0.5	7	0.3	0.2	0.2	19	0.5	0.4	0.4
Social exclusion	38	6	0.3	0.2	0.2	4	0.3	0.2	0.2	6	0.3	0.2	0.2
Stigma experience	24	7	0.3	0.2	0.2	6	0.1	0.1	0.1	6	0.1	0.1	0.1
COMBINED		30	0.4	0.3	0.3	16	0.1	0.1	0.1	28	0.2	0.2	0.2
ATTAINABLEc		95	6.6	5.2	4.4	90	4.2	3.3	2.7	95	6.4	5.1	4.3
APCd		5%	0.3	0.3	0.2	5%	0.2	0.2	0.1	5%	0.3	0.3	0.2

ES- effectiveness of existing interventions at reducing the rate of risk factors. PAF–population attributable fraction; LYG–life years gained at specific ages from reducing a cause weighted by the ES

a Combined = comorbidity adjusted gain in life expectancy for specific determinants

b Lithium is prescribed for bipolar disorder, mainly. Clozapine has been adjusted to reflect that only a third of patients are eligible for prescribing

c Attainable = potential gain in life expectancy considering the combined effect of multiple risk factors

d APC–annual percentage change, together with amount of annual change (rounded data) in life expectancy.

Healthcare systems determinants

Table 2 illustrates that ineffective access to healthcare resources accounted for a larger proportion of all-cause mortality within schizophrenia and schizoaffective disorders (14%) relative to bipolar disorders (5%). Allowing for the effectiveness of current interventions, the potential gain in life expectancy at birth by tackling inadequate healthcare access was three (bipolar disorders) or seven (schizophrenia and schizoaffective disorders) months. The figures in Table 2 have been weighted to reflect the fact that only 30% of people with schizophrenia (treatment-resistant patients) are currently prescribed clozapine. Consequently, improved clozapine prescribing among eligible patients with schizophrenia and improved lithium prescribing within bipolar disorder patients could lead to potential gains in life expectancy at birth of around three months or one year, respectively. Additional gain in life expectancy at age 50 associated with clozapine was two months (schizophrenia) and eight months for lithium (bipolar disorders). Under an additive scenario, improving healthcare practices and processes could increase life expectancy at birth by seven or eight months within the SMI population.

Social determinants

Within social determinants, social deprivation accounted for the largest proportion of avoidable mortality within schizophrenia (21%), schizoaffective disorders (19%), and bipolar disorders (7%). Tackling social deprivation-related inequalities within the SMI population has the potential to increase the life expectancy at birth by on average three (bipolar disorders), five (schizoaffective disorders), or seven (schizophrenia) months. Under an additive scenario, tackling social determinants of all-cause mortality within SMI may increase life expectancy at birth on average by one (bipolar disorders), two (schizoaffective disorders), or four (schizophrenia) months. The potential gain in life expectancy at ages 50 and 65 years from tackling social deprivation was one (bipolar) to three (schizophrenia) months.

When estimated collectively, behavioural habits, healthcare practices, and social determinants accounted for 90% (bipolar disorders) and 95% (schizophrenia and schizoaffective disorders) of all-cause mortality. Allowing for the effectiveness of existing interventions, tackling these modifiable risk factors would lead to a potential gain in life expectancy at birth of four years within bipolar disorders, six years within schizoaffective disorders, or seven years within schizophrenia. Assuming similar APC in life expectancy within men and women with schizophrenia, the estimated life expectancy gain would gradually narrow the current life expectancy gap on average by three years for men and four years among women during a 20-year period (Fig 1). Sensitivity analyses using the average five-year mortality rate produced only marginally lower estimates for gain in life expectancy across the life course validating the robustness of our findings (S2 Table in S1 File).

Fig 1

Trends in life expectancy gain at birth and projected gap (LE Gap) between the general population (GP) and the schizophrenia (SMI) population over a two-decade period.

Discussion

The present study employed a multicontextual approach to quantify potential gains in life expectancy from tackling modifiable causes of premature mortality within specific SMI subgroups. The present study modeling approach relied on two assumptions that (a) the estimated average effectiveness of interventions represented all studies and countries, and (b) that mortality rates of UK adults with SMI applied universally. Accounting for the combined effects of multiple modifiable risk factors, our findings indicated that an attainable target for gain in life expectancy at birth among people with SMI would be around six and a half years for schizophrenia and schizoaffective disorders, and around four years within bipolar disorders. When projected during a 20-year period, these estimates translated into an average annual percentage change in life expectancy at birth of between two to three months (5%). Assuming similar gain in life expectancy at birth among men and women with SMI and using national forecasts for the UK general population, we projected 24% (men) to 28% (women) narrowing of the mortality gap between the general and the SMI populations over a 20-year target. We also estimated that tackling modifiable risk factors for natural causes of death could extend life expectancy at age 50 years three (bipolar disorders) or five (schizophrenia or schizoaffective disorders) years. In addition, tackling modifiable risk factors could extend life expectancy at age 65 years by three (bipolar disorders) or four (schizophrenia and schizoaffective disorders) years.

The implementation of lifestyle interventions depends also on achieving increased acceptance and participation rates on the part of patients. SMIs are commonly associated with impairment in patients’ awareness about the implications of unhealthy behaviours or treatment non-adherence to their health and wellbeing. Poor insight into the illness and its treatment, may challenge clinicians’ capacity to convince SMI patients to adopt healthy behaviours or adhere to treatment. These suggestions are supported by recent evidence documenting that people with SMI failed to manifest similar reduction in mortality rates due to natural causes observed in the general populations over the past decades.[46] Future studies are needed to identify how best to enhance SMI patients’ awareness about the benefits of treatment compliance and lifestyle changes to their health and quality of life.

Smoking emerged as the best single modifiable candidate for increasing life expectancy within schizophrenia (2.4 years), while sedentary behaviour (1.2 years) appeared to be the best single modifiable candidate for increasing life expectancy within bipolar disorder. Yet, lifestyle behaviours within the SMI population continue to be marginalised and poorly integrated into care pathways.[47] For instance, health care professionals were often reluctant to engage in smoking cessation behaviours[48] and smoking cessation services may be less accessible to people with SMI.[49] On the positive side, however, people with SMI appeared equally motivated to want to change negative lifestyle behaviours [50]–though interventions may be less effective.[51] A recent trial illustrated, for instance, that a bespoke smoking cessation intervention embedded in routine mental health care settings[52] was associated with a 56% greater reduction in smoking rates compared to usual care within people with schizophrenia and bipolar disorder. Other studies suggested, however, lower rates of smoking cessation.[53] Regardless of these variation, such findings lend support to our proposed gain in life expectancy within the SMI population, if the effectiveness of current lifestyle interventions can be maintained or improved in the long-term.

Despite valid concerns[54-58] about a potential association between antipsychotic drugs with increased risk of metabolic disorders, previous studies revealed a potential for clozapine and lithium to reduce premature mortality among people with schizophrenia and bipolar disorders.[59-61] This may seem counter-intuitive given that clozapine is prone to cause obesity, diabetes, cardiomyopathy[58] and, rarely, agranulocytosis, as well as being reserved for the most severe, treatment resistant group. Also, lithium has a narrow therapeutic range, and is potentially associated with renal impairment or failure.[62,63] Prior research suggest that some of these side-effects may be offset by the evidence for reduced all-cause mortality. For instance, studies have shown up to 50% reduction in all-cause mortality among people with schizophrenia receiving clozapine compared to those never prescribed the drug.[64] The mechanisms through which these drugs have their beneficial effects may be a direct consequence of treatment on mental health symptoms, or on service related factors such as continued monitoring and continuity of care which is necessitated by the routine monitoring involved. While effective, clozapine and lithium are not always accepted by patients and require expertise and experience in order to be prescribed safely. Thus, there are implementation challenges to any policy highlighting their wider use.

The findings also underline the potential for additional gain in life expectancy through facilitating timely access to healthcare resources and improved prescribing practices. Poor access to healthcare resources contribute to increased mortality rates among people with SMI,[65] though UK estimates of reduced life expectancy in people with SMI are less dramatic than those arising from other healthcare systems[66] (possibly reflecting the universal access offered by the NHS). A confluence of clinician (e.g. insufficient assessment, poor communication, suboptimal prescribing habits), service providers (e.g. poor care coordination, insufficient funds), and patient (e.g. low motivation leading to poor adherence to treatment, difficulties in understanding health care advice) factors may account for the poor access to and quality of health care among people with SMI. These factors all contribute to the reduced life expectancy among people with SMI, [67] as documented in this study. Recent research has confirmed disparities in access to specialist medical interventions (e.g. revascularisation) following a major health event among people with SMI.[68] Diagnostic delays related to physical comorbidities [69, 70] may also interfere with timely access to healthcare resources and effective treatment, increasing the risk of premature mortality. Our study findings suggest that tackling patient and health system barriers to accessing relevant healthcare resources and medical interventions would extend life expectancy by four to eight months across the life course within the broader SMI population.

Stigma and social isolation may influence mortality outcomes indirectly by impacting on lifestyle behaviours, treatment continuation or adherence, and the timely utilisation of healthcare resources.[71]. Stigma associated with SMI has been identified as a major barrier to accessing treatment and health care resources, and it occurs at all levels including healthcare professionals, society, and individuals.[72]. Similarly, socioeconomic inequalities in mortality outcomes among people with SMI are likely to reflect inequalities in access to healthcare resources, low health care-seeking behaviour, poorer knowledge of physical health symptoms and risk factors, and motivational challenges to accessing healthcare support. Our findings implied that the implementation of wider initiatives to tackle poverty and stigma would improve life expectancy among people with SMI by around 10%.

Strengths and limitations of this study

To our knowledge this is the first study to consider both the population attribution fraction and the effectiveness of interventions in calculating the expected gain in life expectancy over the life course within specific SMI populations. Another major strength is the use of a multicontextual approach to estimates potential gain in life expectancy over the lifecourse across specific SMI subgroups. Our estimation model shares limitations common to most projections of future gain in life expectancy, such as proportionality hypothesis (e.g. constant mortality rates and intervention effectiveness over time). We have partially addressed this concern by adopting short-term projections and conducting sensitivity analyses averaging mortality rates over a five-year period. Also, our estimates for treatment impact on life expectancy gains have been constrained to two drugs with the greatest evidence on harms and benefits within the SMI populations. Thus, our findings may not generalise to patients prescribed other antipsychotic or antidepressant drugs, or to polypharmacy. Given limited resources, it is not always possible to conduct studies on all issues in all settings prior to making a policy decision. This was also the case with the present study, where no single country collected relevant data on all study parameters estimated in the models. This concern places constraints about the generalizability of present study findings across different countries with varying healthcare and socioeconomic contexts. Ultimately, it is the responsibility of decision-makers to consider the extent to which evidence from one setting (UK) is transferrable to a different setting. Regardless, the framework set out in this study identified critical gaps in the current evidence base that may encourage future research and developments into modeling gains in life expectancy from addressing modifiable determinants of premature mortality among people with SMI. There were suggestions that PAF formula may overestimate excess burden of all-cause mortality due to smoking, [73] and this concern applies to our study estimates including the high collective estimates. Further, when confounders exist and one does not correct for this, the PAF is likely to be influenced. Moreover, residual confounding also exists and therefore the PAF suffers from overestimation. PAF estimates tend to vary, however, across populations, over time and with different ranking of common mortality causes.[74] It also provides healthcare providers and policy makers with a useful tool to interpret the excess mortality due to specific factors among people with SMI.[74] The healthcare is a more complex system than conceptualised here, and our findings may underestimate the potential gain in life expectancy from healthcare-based interventions within the broader SMI population. Interventions at any level of the healthcare system, however, will be reflected in changes across other parts of the system including the healthcare characteristics considered in this study. Moreover, while the gap in life expectancy among men and women with SMI is similar (i.e. 13 and 12, respectively, years) future studies assessing possible differential gain in life expectancy among men and women with specific SMIs are warranted. When translating evidence from well-controlled trials into clinical practice, the dilution of the intervention effect is common. This concern is exacerbated by SMI patients representing a challenging group to treat, for several reasons (e.g. reduced insights into their condition, concerns around drugs side-effects). These concerns imply the need for greater efforts to accomplish the effects sizes used in our modeling approach. Several suggestions were put forward (e.g. improved mental and study physical care coordination, address patients’ resistance to treatment or screening uptake) how this could be achieved, yet, evidence to confirm the feasibility of these proposals is currently lacking. Finally, our study estimates relied on UK-based data for mortality rates, while data on prevalence of modifiable risk factors and effectiveness of interventions coming mainly from developed countries. These estimates may not be directly transferrable to other countries with a different care system and epidemiology of risk factors social circumstances, or health care provision. We have aimed to moderate this concern by relying, whenever possible, on meta-analysis findings that accounts for variation in different study designs and methodology.

Conclusion

Our study findings indicated that addressing unhealthy behaviours, suboptimal use of healthcare resources, and poor life circumstances have the potential prolong life expectancy at birth by four and six years within bipolar or schizophrenia disorders. If our study estimates are translatable into routine practice, we would expect around 24% (men) to 28% (women) narrowing in the life expectancy gap between the SMI and the general population from tackling modifiable risk factors. Given similar or greater life expectancy gaps in other Western countries these results suggest that at least comparable improvements in the longevity of people with SMI can be achieved elsewhere. Achieving these projections requires multisectoral approach to ensure that the complexity of SMI disorders are addressed at individual, clinical, and societal level.[75] The study findings need to be interpreted cautiously since translation of clinical trials evidence into routine care it is often challenging and ineffective.

(DOCX)

Click here for additional data file.

18 Nov 2019

PONE-D-19-27607

Potential gains in life expectancy from reducing amenable mortality among people diagnosed with serious mental illness

PLOS ONE

Dear Dr. Dregan,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

We would appreciate receiving your revised manuscript by Jan 02 2020 11:59PM. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

We look forward to receiving your revised manuscript.

Kind regards,

Sinan Guloksuz, M.D., Ph.D.

Academic Editor

PLOS ONE

Journal Requirements:

1. When submitting your revision, we need you to address these additional requirements.

Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

http://www.journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and http://www.journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Please include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: http://journals.plos.org/plosone/s/supporting-information.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: No

Reviewer #2: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: No

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: No

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: This is a study using population attributable risk factors to estimate potential gains in life expectancy in serious mental illnesses (schizophrenia, schizoaffective disorder and bipolar disorder).

1. This article is based on two assumptions: 1) the average effectiveness of a treatment intervention represent all studies and all countries, and 2) changes in mortality rate from adults with SMI in the United Kingdom applies to all countries.

2. I do not see the estimations of the effectiveness of the interventions for the various modifiable risks. Please add them to Table 1.

3. I am a clinician, so the idea that increasing the effectiveness of interventions related to lifestyle factors can increase life expectancy makes sense from a medical point of view. The only problem that I see is there is no acknowledgement that some of the lack of implementation of these interventions is due to the lack of cooperation of patients with SMI. These SMIs are associated with impairment in insight. You cannot force patients to do what they do not want to do. In summary, in an ideal system with an ideal physician or ideal health care, some of the lack of application of effective interventions may still be due to lack of interest or participation on the part of patients.

4. The idea of modifying healthcare factors and social factors is beyond what physicians do. These appear to be interventions at the political level. To mix healthcare interventions and politics does not seem wise to me. It appears to mix apples and oranges.

5. For a clinician like me, this article appears to be an exercise in mathematical modeling with limited applicability to the real world. Moreover, the title should reflect the fact that this mortality reduction applies only to the United Kingdom. It is laughable to think that these estimations have any value in countries with very different life expectancies or very different health systems. It also may not hurt to explain somehow in the title that this study was done using average rates of effectiveness from studies in multiple countries. Again, it is not likely that interventions such as reducing smoking or obesity in people with SMI would apply homogeneously across Western countries. Different Western countries are at different stages of change in the general population regarding these factors and the application of these interventions. If the data on interventions also includes other countries, such as those from Asia, that makes no sense at all. In summary, the data on interventions should come from the United Kingdom if you want to play to life expectancy in the United Kingdom. If the data on interventions comes from countries with very different life expectancies and different health systems, I do not see how this data can be used in a model based on the life expectancy of people with SMI in the United Kingdom.

6. The Discussion does not reflect awareness of the limitations of mathematical modeling.

7. “A recent trial illustrated, for instance, that a bespoke smoking cessation intervention embedded in routine mental health care settings (51) was associated with a 56% greater reduction in smoking rates compared to usual care within people with schizophrenia and bipolar disorder. This finding lends support to our proposed gain in life expectancy within the SMI population, if the effectiveness of current lifestyle interventions can be maintained or improved in the long-term.” This paragraph is a serious misrepresentation of that study and its follow-up study. Reference 51 is a pilot study that reports 12-month smoking cessation rates of 69% in 51 controls and 72% among 46 in the intervention group. Then there is a later study Gilbody S, Peckham E, Bailey D, Arundel C, Heron P, Crosland S, Fairhurst C, Hewitt C, Li J, Parrott S, Bradshaw T, Horspool M, Hughes E, Hughes T, Ker S, Leahy M, McCloud T, Osborn D, Reilly J, Steare T, Ballantyne E, Bidwell P, Bonner S, Brennan D, Callen T, Carey A, Colbeck C, Coton D, Donaldson E, Evans K, Herlihy H, Khan W, Nyathi L, Nyamadzawo E, Oldknow H, Phiri P, Rathod S, Rea J, Romain-Hooper CB, Smith K, Stribling A, Vickers C. Smoking cessation for people with severe mental illness (SCIMITAR+): a pragmatic randomised controlled trial. Lancet Psychiatry. 2019 May;6(5):379-390. doi: 10.1016/S2215-0366(19)30047-1.Epub 2019 Apr 8. PubMed PMID: 30975539; PubMed Central PMCID: PMC6546931. In this study, “The incidence of quitting at 6 months shows that smoking cessation can be achieved, but the waning of this effect by 12 months means more effort is needed for sustained quitting.” In summary, unfortunately, at 12 months the effect disappeared.

The most pessimistic interpretation is that we do not have any practical intervention for providing long-term smoking cessation in large groups of these patients. If you have any published intervention that has demonstrated that, please quote it. As indicated before, the pilot study quoted by the authors led to an unsuccessful trial. In my experience and through review of the long-term data in my state, some patients are able to stop on their own but, unfortunately, we as the health providers are not being very helpful.

8. Please delete the statement, “Our study findings corroborate with earlier evidence that effective mental healthcare would, in and of itself, be a potent means of reducing premature mortality by

addressing underlying symptoms and social problems arising from SMI.” This is not an independent study. You are making multiple assumptions using prior literature. It is not surprising that a mathematical model using prior literature supports the prior literature.

9. I think that clozapine and lithium are excellent drugs and should be used much more frequently. Many times, patients do not want to use them and you cannot force them to take them. They are generic drugs that are not promoted by pharmaceutical companies. Moreover, they are mainly started and mainly managed by psychiatrists due to their complex pharmacology. My psychiatry residents do not know how to prescribe them since most of the attendings in my academic department do not use them. Thus, I am not optimistic that in the future they will be prescribed more frequently, at least not in my state in the US.

10. There are many studies on the barriers involved in the use of clozapine. “Verdoux H, Quiles C, Bachmann CJ, Siskind D. Prescriber and institutional barriers and facilitators of clozapine use: A systematic review. Schizophr Res. 2018 Nov;201:10-19. doi: 10.1016/j.schres.2018.05.046. Epub 2018 Jun 4. PubMed PMID: 29880453.” The truth is that people like me, who consider themselves experts on clozapine, appear to be incompetent in overcoming these barriers where they practice. It would be helpful if the authors would teach us how to increase the use of clozapine or lithium. They appear to know things that we do not know.

11. The Limitations do not reflect any of the prior limitations of using data on the effectiveness of intervention from many countries and then applying it to the life expectancy of people with SMI in the United Kingdom and then trying to generalize it to the whole world.

12. There is no attempt to consider the lack of cooperation of patients and physicians in improving the dismal situation surrounding the life expectancy of people with SMI. I work as a consultant in the public system of a state in the US. The first problem for me in implementing basic interventions such as increasing the use of clozapine and lithium is that some clinicians do not want to deal with their complications and, in the case of clozapine, with much more paperwork. Once I am dealing with convinced and trained clinicians, they need to convince each individual patient and their families. Nobody is paying for advertisements for these two generic drugs. Pharmaceutical companies support other antipsychotics and other mood stabilizers that are competing with clozapine and lithium. I would like to live in the same mathematical universe as the authors and believe that in my state these two drugs will be more widely prescribed because it is the right thing to do.

13. Please understand that I do not deny that the authors have very good intentions, but estimating the effect of interventions without considering the barriers does not appear very useful in the real world. On the other hand, I acknowledge that modeling the barriers to implementation will not be easy.

Reviewer #2: This is an important manuscript to enhance implementation of effective treatments for modifiable risk factors. The authors describe an important effort to summarize literature and calculate with the numbers from previous studies.

However, I have some points to consider:

Abstract:

- please mention the timeframe of the data that was used to calculate your results.

-conclusions: These % are under ideal circumstances and without the limitation of overestimation which often comes with PAFs, please rephrase this cautiously.

Introduction:

-these diseases are party attributable,. Obesity can also be caused by olanzapine and clozapine and therefore these factors might not be as easily tackled as we might wish.

Methods:

-Why not update the literature beyond 2018? For example, a recent study found an increasing number of years life lost https://www.ncbi.nlm.nih.gov/pubmed/30446270

Also, the results of the scimitar trial regarding smoking are recently published which gives important nuances in how hard treatment is in these groups.

-Again, modifiable risk factors: treatment with certain antipsychotics induces the risk of cardiovasculair disease (see De Hert 2012, nature reviews) and therefore for example is less modifiable than we hope. This should be at least mentioned if one cannot correct for this in some way in the analyses.

-The use of PAFs and formula has several limitations and overestimation is likely to occur

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4339639/

Please elaborate on the choice for the PAF and why this particular formal was chosen, here and/or in the limitations section of the manuscript.

- 36% for smoking cessation is likely to be an overestimation (see scimitair results british journal of psychiatry Gilbody et al.)

Results:

Is it possible to correct for the interaction of the factors in Table 1 and the RRs for mortality?

Healthcare system determinants: What about the increased risk for adverse effects that come with lithium and antipsychotics, does this balance out against the gains?

Collective estimates: 90% seems high, is there a possibility of overestimation?

Discussion

Indeed, standard interventions are less effective, more effort Is needed to accomplish similar effect sizes. SMI patients are a harder to treat population, please elaborate on this and how we can improve our interventions.

Please add the long-term meta-analysis on antipsychotics/clozapine and mortality to the litertarue (ref56-58)

Limitations: When confounders exist and one does not correct for this, the PAF is likely to be influenced. Moreover, residual confounding also exists and therefore the PAF suffers from overestimation. Please mention this here.

What are the limitations regarding generalizability to non-western countries?

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: Yes: Jentien Vermeulen, MD PhD

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files to be viewed.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org. Please note that Supporting Information files do not need this step.

12 Feb 2020

REVIEWER #1 COMMENTS:

1. This article is based on two assumptions: 1) the average effectiveness of a treatment intervention represent all studies and all countries, and 2) changes in mortality rate from adults with SMI in the United Kingdom applies to all countries.

R: The Reviewer makes a valid observation about our study methodological assumption. We have added the following sentences in the Discussion section on Page 16: “The present study modeling approach relied on two assumptions that (a) the estimated average effectiveness of interventions represented all studies and countries, and (b) that mortality rates of UK adults with SMI applied universally.”

2. I do not see the estimations of the effectiveness of the interventions for the various modifiable risks. Please add them to Table 1.

R: Please note that the estimations of the effectiveness of the interventions for the modifiable risks are included in Table 2, second column (ES).

3. I am a clinician, so the idea that increasing the effectiveness of interventions related to lifestyle factors can increase life expectancy makes sense from a medical point of view. The only problem that I see is there is no acknowledgement that some of the lack of implementation of these interventions is due to the lack of cooperation of patients with SMI. These SMIs are associated with impairment in insight. You cannot force patients to do what they do not want to do. In summary, in an ideal system with an ideal physician or ideal health care, some of the lack of application of effective interventions may still be due to lack of interest or participation on the part of patients.

R: We are now stating in the Discussion section on Pag 15, that: “The implementation of lifestyle interventions depends also on achieving increased acceptance and participation rates on the part of patients. SMIs are commonly associated with impairment in patients’ awareness about the implications of unhealthy behaviours or treatment non-adherence to their health and wellbeing. Poor insight into the illness and its treatment, may challenge clinicians’ capacity to convince SMI patients to adopt healthy behaviours or adhere to treatment. Future studies are needed to identify how best to enhance SMI patients’ awareness about the benefits of treatment compliance and lifestyle changes to their health and quality of life.”

Please also note that on Page 18 we did mention that several factors contribute to challenges in implementing lifestyle-oriented interventions: “A confluence of clinician (e.g. insufficient assessment, poor communication, suboptimal prescribing habits), service providers (e.g. poor care coordination, insufficient funds), and patient (e.g. low motivation leading to poor adherence to treatment, difficulties in understanding health care advice) factors may account for the poor access to and quality of health care among people with SMI.”

4. The idea of modifying healthcare factors and social factors is beyond what physicians do. These appear to be interventions at the political level. To mix healthcare interventions and politics does not seem wise to me. It appears to mix apples and oranges.

R: While we agree that social factors are beyond physicians’ usual responsibilities, we respectfully disagree with the Reviewer suggestion that mixing healthcare and politics are necessarily opposites. To bring about sustainable health and life-expectancy benefits to patients with SMI requires multisystem interventions involving multiple stakeholders, including patients, clinicians, health care systems, policy makers, researchers, and employers. In this respect, we did state throughout the Discussion section that multiple stakeholders need to be involved in improving life expectancy in people with SMI, beyond physicians. For instance, on Page 18 we have stated that: “Our study findings suggest that tackling patient and health system barriers to accessing relevant healthcare resources and medical interventions would extend life expectancy by four to eight months across the life course within the broader SMI population.” On Page 19 we have acknowledged that: “Our findings implied that the implementation of wider initiatives to tackle poverty and stigma would improve life expectancy among people with SMI by around 10%.” Further, the last sentence of the Conclusion stated: “Achieving these projections requires multisectoral approach to ensure that the complexity of SMI disorders are addressed at individual, clinical, and societal level.”

5. For a clinician like me, this article appears to be an exercise in mathematical modeling with limited applicability to the real world. Moreover, the title should reflect the fact that this mortality reduction applies only to the United Kingdom. It is laughable to think that these estimations have any value in countries with very different life expectancies or very different health systems. It also may not hurt to explain somehow in the title that this study was done using average rates of effectiveness from studies in multiple countries. Again, it is not likely that interventions such as reducing smoking or obesity in people with SMI would apply homogeneously across Western countries. Different Western countries are at different stages of change in the general population regarding these factors and the application of these interventions. If the data on interventions also includes other countries, such as those from Asia, that makes no sense at all. In summary, the data on interventions should come from the United Kingdom if you want to play to life expectancy in the United Kingdom. If the data on interventions comes from countries with very different life expectancies and different health systems, I do not see how this data can be used in a model based on the life expectancy of people with SMI in the United Kingdom.

R: The Reviewer makes a valid point about the challenges associated with applicability and transferability of findings across different healthcare systems. To address this concern we have relied, whenever possible, on data from meta-analyses. We have included the following statement on Page 19: “Given limited resources, it is not always possible to conduct studies on all issues in all settings prior to making a policy decision. This was also the case with the present study, where no single country collected relevant data on all study parameters estimated in the models. This concern places constraints about the generalizability of present study findings across different countries with varying healthcare and socioeconomic contexts. Ultimately, it is the responsibility of decision-makers to consider the extent to which evidence from one setting (UK) is transferrable to a different setting. Regardless, the framework set out in this study identified critical gaps in the current evidence base that may encourage future research and developments into modeling gains in life expectancy from addressing modifiable determinants of premature mortality among people with SMI.” We have also changed the study title to: “Potential gains in life expectancy from reducing amenable mortality among people diagnosed with serious mental illness in the United Kingdom.”

6. The Discussion does not reflect awareness of the limitations of mathematical modeling.

R: Please see our reply to Reviewer 2, Points, 5, 7, and 13 which discuss the concerns around PAF estimates reliability and the lack of modeling for interaction effects.

7. “A recent trial illustrated, for instance, that a bespoke smoking cessation intervention embedded in routine mental health care settings (51) was associated with a 56% greater reduction in smoking rates compared to usual care within people with schizophrenia and bipolar disorder. This finding lends support to our proposed gain in life expectancy within the SMI population, if the effectiveness of current lifestyle interventions can be maintained or improved in the long-term.” This paragraph is a serious misrepresentation of that study and its follow-up study. Reference 51 is a pilot study that reports 12-month smoking cessation rates of 69% in 51 controls and 72% among 46 in the intervention group. Then there is a later study Gilbody S, Peckham E, Bailey D, Arundel C, Heron P, Crosland S, Fairhurst C, Hewitt C, Li J, Parrott S, Bradshaw T, Horspool M, Hughes E, Hughes T, Ker S, Leahy M, McCloud T, Osborn D, Reilly J, Steare T, Ballantyne E, Bidwell P, Bonner S, Brennan D, Callen T, Carey A, Colbeck C, Coton D, Donaldson E, Evans K, Herlihy H, Khan W, Nyathi L, Nyamadzawo E, Oldknow H, Phiri P, Rathod S, Rea J, Romain-Hooper CB, Smith K, Stribling A, Vickers C. Smoking cessation for people with severe mental illness (SCIMITAR+): a pragmatic randomised controlled trial. Lancet Psychiatry. 2019 May;6(5):379-390. doi: 10.1016/S2215-0366(19)30047-1.Epub 2019 Apr 8. PubMed PMID: 30975539; PubMed Central PMCID: PMC6546931. In this study, “The incidence of quitting at 6 months shows that smoking cessation can be achieved, but the waning of this effect by 12 months means more effort is needed for sustained quitting.” In summary, unfortunately, at 12 months the effect disappeared.

The most pessimistic interpretation is that we do not have any practical intervention for providing long-term smoking cessation in large groups of these patients. If you have any published intervention that has demonstrated that, please quote it. As indicated before, the pilot study quoted by the authors led to an unsuccessful trial. In my experience and through review of the long-term data in my state, some patients are able to stop on their own but, unfortunately, we as the health providers are not being very helpful.

R: We thank you the reviewer for this valuable insight. We have responded on a similar concern raised by Reviewer 2, Point 6 related to the Gilbody et al. Scimitar trial which has been added as a reference in the study.

8. Please delete the statement, “Our study findings corroborate with earlier evidence that effective mental healthcare would, in and of itself, be a potent means of reducing premature mortality by

addressing underlying symptoms and social problems arising from SMI.” This is not an independent study. You are making multiple assumptions using prior literature. It is not surprising that a mathematical model using prior literature supports the prior literature.

R: As per Reviewer comment, we have now excluded the statement from the manuscript.

9. I think that clozapine and lithium are excellent drugs and should be used much more frequently. Many times, patients do not want to use them and you cannot force them to take them. They are generic drugs that are not promoted by pharmaceutical companies. Moreover, they are mainly started and mainly managed by psychiatrists due to their complex pharmacology. My psychiatry residents do not know how to prescribe them since most of the attendings in my academic department do not use them. Thus, I am not optimistic that in the future they will be prescribed more frequently, at least not in my state in the US.

R: We share Reviewer’ opinion on clozapine and lithium, however these concerns are not unique to these drugs and innovation in medicine would scarcely progress if we adopted such a pessimistic stance. We have revised Page 18 to state that: “While effective, clozapine and lithium are not always accepted by patients and require expertise and experience in order to be prescribed safely. Thus, there are implementation challenges to any policy highlighting their wider use.”

10. There are many studies on the barriers involved in the use of clozapine. “Verdoux H, Quiles C, Bachmann CJ, Siskind D. Prescriber and institutional barriers and facilitators of clozapine use: A systematic review. Schizophr Res. 2018 Nov;201:10-19. doi: 10.1016/j.schres.2018.05.046. Epub 2018 Jun 4. PubMed PMID: 29880453.” The truth is that people like me, who consider themselves experts on clozapine, appear to be incompetent in overcoming these barriers where they practice. It would be helpful if the authors would teach us how to increase the use of clozapine or lithium. They appear to know things that we do not know.

R: We have emphasized throughout the manuscript the about the need for multisectorial and multifaceted efforts to address concerns around the gap in life expectancy among people with SMI. For instance, on Page 18 we have stated: “A confluence of clinician (e.g. insufficient assessment, poor communication, suboptimal prescribing habits), service providers (e.g. poor care coordination, insufficient funds), and patient (e.g. low motivation leading to poor adherence to treatment, difficulties in understanding health care advice) factors may account for the poor access to and quality of health care among people with SMI. These factors all contribute to the reduced life expectancy among people with SMI, (64) as documented in this study.” On Page 21 we have stated: “Our study findings indicated that addressing unhealthy behaviours, suboptimal use of healthcare resources, and poor life circumstances have the potential prolong life expectancy at birth by four and six years within bipolar or schizophrenia disorders.”

11. The Limitations do not reflect any of the prior limitations of using data on the effectiveness of intervention from many countries and then applying it to the life expectancy of people with SMI in the United Kingdom and then trying to generalize it to the whole world.

R: We have included the following sentences on Page 21 of the Limitations sections: “The present study estimates and findings relied on UK-related data for mortality rates, while data on prevalence of modifiable risk factors and effectiveness of interventions coming mainly from developed countries. These estimates may not be directly transferrable to other countries with a different care system and epidemiology of risk factors social circumstances, or health care provision. We have aimed to moderate this concern by relying, whenever possible, on meta-analysis findings that accounts for variation in different study designs and methodology.”

12. There is no attempt to consider the lack of cooperation of patients and physicians in improving the dismal situation surrounding the life expectancy of people with SMI. I work as a consultant in the public system of a state in the US. The first problem for me in implementing basic interventions such as increasing the use of clozapine and lithium is that some clinicians do not want to deal with their complications and, in the case of clozapine, with much more paperwork. Once I am dealing with convinced and trained clinicians, they need to convince each individual patient and their families. Nobody is paying for advertisements for these two generic drugs. Pharmaceutical companies support other antipsychotics and other mood stabilizers that are competing with clozapine and lithium. I would like to live in the same mathematical universe as the authors and believe that in my state these two drugs will be more widely prescribed because it is the right thing to do.

R: The Reviewer may have misunderstood the purpose of our paper. We are testing out areas where there might be tractable targets for intervention to address the health disparity (increased mortality in SMI). We do not pretend that the interventions proposed can easily be implemented, but we do think that it is reasonable to model the potential gains where one might have the greatest traction. We share the Reviewer’s concerns around dealing with complex patients and challenges around non-adherence with treatment/control recommendations, which we consider are highly context specific.

13. Please understand that I do not deny that the authors have very good intentions, but estimating the effect of interventions without considering the barriers does not appear very useful in the real world. On the other hand, I acknowledge that modeling the barriers to implementation will not be easy.

R: Thank you, this point is addressed above under point 12. We naturally understand the concern of the Reviewer and have added the following sentence in the Conclusion section, Page 21: “The study findings need to be interpreted cautiously since translation of clinical trials evidence into routine care it is often challenging and ineffective.”

Reviewer #2 (Comments to Author):

Summary:

This is an important manuscript to enhance implementation of effective treatments for modifiable risk factors. The authors describe an important effort to summarize literature and calculate with the numbers from previous studies.

R: Thank you for the positive evaluation of our research.

Specific Comments:

1. Abstract:

- please mention the timeframe of the data that was used to calculate your results.

-conclusions: These % are under ideal circumstances and without the limitation of overestimation which often comes with PAFs, please rephrase this cautiously.

R: We are now clarifying that the timeframe of the data as: “The predicted estimates were based on mortality rates for year 2014-2015.“ We have now included the following text in the Abstract: “These projections represent ideal circumstances and without the limitation of overestimation which often comes with PAFs.“

2. Introduction:

-these diseases are party attributable,. Obesity can also be caused by olanzapine and clozapine and therefore these factors might not be as easily tackled as we might wish.

R: Clarified on Page 3, Introduction that these diseases were partly attributable.

3. Methods:

-Why not update the literature beyond 2018? For example, a recent study found an increasing number of years life lost https://www.ncbi.nlm.nih.gov/pubmed/30446270. Also, the results of the scimitar trial regarding smoking are recently published which gives important nuances in how hard treatment is in these groups.

R: We thank the Reviewer for the helpful information. We have added the following text on Page 18: “These suggestions are supported by recent evidence documenting that people with SMI failed to manifest similar reduction in mortality rates due to natural causes observed in the general populations over the past decades.(46)”

4. -Again, modifiable risk factors: treatment with certain antipsychotics induces the risk of cardiovasculair disease (see De Hert 2012, nature reviews) and therefore for example is less modifiable than we hope. This should be at least mentioned if one cannot correct for this in some way in the analyses.

R: Please note that on Page 18 we did mention that antipsychotics have adverse effects on cardiometabolic disorders. We have now added the De Hert et al. (2012) reference on Page 18:” Despite valid concerns(53-57) about a potential association between antipsychotic drugs with increased risk of metabolic disorders, …”. We have also added the De Hert et al. reference on Page 19 where have stated that:” This may seem counter-intuitive given that clozapine is prone to cause obesity, diabetes, cardiomyopathy(57) and…”.

5. -The use of PAFs and formula has several limitations and overestimation is likely to occur

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4339639/ Please elaborate on the choice for the PAF and why this particular formal was chosen, here and/or in the limitations section of the manuscript.

R: We are now clarifying on Page 21 that: “There are suggestions that PAF formula may overestimate excess burden of all-cause mortality due to smoking, (72) and this concern applies to our study estimates as well. PAF estimates tend to vary, however, across populations, over time and with different ranking of common mortality causes.(73) It also provides healthcare providers and policy makers with a useful tool to interpret the excess mortality due to specific factors among people with SMI.(73)”

6. - 36% for smoking cessation is likely to be an overestimation (see scimitair results british journal of psychiatry Gilbody et al.)

R: We have now clarified on Page 17 that: “Other studies suggested, however, lower rates of smoking cessation.(53) Regardless of these variation, such…”

7. Results:

Is it possible to correct for the interaction of the factors in Table 1 and the RRs for mortality?

R: Please note that on Page 6 we did state: “To allow for the overlap between different modifiable risk factors in influencing all-cause mortality, we have also estimated the combined impact of multiple modifiable risk factors within each group of determinants (e.g. behavioural, healthcare, social) and across all modifiable risk factors… While the formula assumes no interaction effects, it ensures that the PAF for the combined contribution of modifiable risk factors to all-cause mortality is not greater than 1.(28) “

8. Healthcare system determinants: What about the increased risk for adverse effects that come with lithium and antipsychotics, does this balance out against the gains?

R: We agree that this a valid point, yet very challenging to estimate in practice. We are not aware of any multi-component trial that has considered the simultaneous effects of healthcare and therapeutic pathways on excess mortality within SMI. We do not think that the evidence suggests that adverse effects of lithium and clozapine balance out the gains – otherwise we would not be suggesting these as potential interventions. As stated on Page 17: “Despite valid concerns(54-58) about a potential association between antipsychotic drugs with increased risk of metabolic disorders, previous studies revealed a potential for clozapine and lithium to reduce premature mortality among people with schizophrenia and bipolar disorders.(59-61).”

9. Collective estimates: 90% seems high, is there a possibility of overestimation?

R: Please note that these estimates are context specific and our study has considered most common modifiable factors supported by the existing literature. We have included the following clarification on Page 20: “…including the high (90%) collective estimates.”

10. Discussion

Indeed, standard interventions are less effective, more effort Is needed to accomplish similar effect sizes. SMI patients are a harder to treat population, please elaborate on this and how we can improve our interventions.

R: We are now saying on Page 21 that: “When translating evidence from well-controlled trials into clinical practice, the dilution of the intervention effect is common. This concern is exacerbated by SMI patients representing a challenging group to treat, for several reasons (e.g. reduced insights into their condition, concerns around drugs side-effects). These concerns imply the need for greater efforts to accomplish the effects sizes used in our modeling approach. Several suggestions were put forward (e.g. improved mental and study physical care coordination, address patients’ resistance to treatment or screening uptake) how this could be achieved, yet, evidence to confirm the feasibility of these proposals is currently lacking.”

11. Please add the long-term meta-analysis on antipsychotics/clozapine and mortality to the litertarue (ref56-58)

R: As per Reviewer suggestion we have now added De Hert et al. meta-analysis study on antipsychotics/clozapine.

12. Limitations: When confounders exist and one does not correct for this, the PAF is likely to be influenced. Moreover, residual confounding also exists and therefore the PAF suffers from overestimation. Please mention this here.

R: We have added the following sentence on Page 20: “Further, when confounders exist and one does not correct for this, the PAF is likely to be influenced. Moreover, residual confounding also exists and therefore the PAF suffers from overestimation.”

13. What are the limitations regarding generalizability to non-western countries?

R. Please see our reply to Reviewer 1, Point 11.

EDITORIAL REQUIREMENTS

1. When submitting your revision, we need you to address these additional requirements.

Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

http://www.journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and http://www.journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

R: Our revised ms meets the PLOS ONE’s style requirements.

2. Please include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: http://journals.plos.org/plosone/s/supporting-information.

Please subject Figure 1 at a minimum resolution of >350 ppi or in one of the following VECTOR formats: .ai, .docx, .emf, .eps, .pdf

R: We have adddressed the requirements for the Supplementary material and have enhanced the resolution of Figure 1.

Statistical Graphs

For Figure 2, please provide a graph output directly from the software used to create it in an editable VECTOR file format, such as .wmf or .eps, or as an Excel graph, if created in Excel. If you provide .pdf files, be sure that these are in VECTOR file formal, not Raster file format. Raster (picture) files, such as .jpg or .tif, output directly or embedded in vector files, are not acceptable. All statistical graphs in accepted manuscripts are recreated in-house.

R: We have saved and attached Figure 2 as an .wmf file

References:

1. Ridker PM, Buring JE, Cook NR, et al. C-reactive protein, the metabolic syndrome, and risk of incident cardiovascular events: an 8-year follow-up of 14 719 initially healthy American women. Circulation 2003;107(3):391-7. doi: 10.1161/01.cir.0000055014.62083.05 [published Online First: 2003/01/29]

2. Sesso HD, Buring JE, Rifai N, et al. C-reactive protein and the risk of developing hypertension. JAMA 2003;290(22):2945-51. doi: 10.1001/jama.290.22.2945 [published Online First: 2003/12/11]

3. Groenwold RH, Sterne JA, Lawlor DA, et al. Sensitivity analysis for the effects of multiple unmeasured confounders. Ann Epidemiol 2016;26(9):605-11. doi: 10.1016/j.annepidem.2016.07.009 [published Online First: 2016/09/01]

4. Yusuf S, Hawken S, Ounpuu S, et al. Effect of potentially modifiable risk factors associated with myocardial infarction in 52 countries (the INTERHEART study): case-control study. Lancet 2004;364(9438):937-52. doi: 10.1016/S0140-6736(04)17018-9 [published Online First: 2004/09/15]

5. Hare DL, Toukhsati SR, Johansson P, et al. Depression and cardiovascular disease: a clinical review. Eur Heart J 2014;35(21):1365-72. doi: 10.1093/eurheartj/eht462 [published Online First: 2013/11/28]

6. Khambaty T, Stewart JC, Gupta SK, et al. Association Between Depressive Disorders and Incident Acute Myocardial Infarction in Human Immunodeficiency Virus-Infected Adults: Veterans Aging Cohort Study. JAMA Cardiol 2016;1(8):929-37. doi: 10.1001/jamacardio.2016.2716 [published Online First: 2016/08/25]

7. Whooley MA. Depression and cardiovascular disease: healing the broken-hearted. JAMA 2006;295(24):2874-81. doi: 10.1001/jama.295.24.2874 [published Online First: 2006/06/29]

8. Alberti KG, Eckel RH, Grundy SM, et al. Harmonizing the metabolic syndrome: a joint interim statement of the International Diabetes Federation Task Force on Epidemiology and Prevention; National Heart, Lung, and Blood Institute; American Heart Association; World Heart Federation; International Atherosclerosis Society; and International Association for the Study of Obesity. Circulation 2009;120(16):1640-5. doi: 10.1161/CIRCULATIONAHA.109.192644 [published Online First: 2009/10/07]

9. Millasseau SC KR, Ritter JM, Chowienczyk PJ. Determination of age‐related increases in large artery stiffness by digital pulse contour analysis. Clinical science (London) 2002;103:371-77.

10. Lewis TT, Sutton-Tyrrell K, Penninx BW, et al. Race, psychosocial factors, and aortic pulse wave velocity: the Health, Aging, and Body Composition Study. J Gerontol A Biol Sci Med Sci 2010;65(10):1079-85. doi: 10.1093/gerona/glq089 [published Online First: 2010/06/05]

11. Tiemeier H, Breteler MM, van Popele NM, et al. Late-life depression is associated with arterial stiffness: a population-based study. J Am Geriatr Soc 2003;51(8):1105-10. [published Online First: 2003/08/02]

12. Seldenrijk A, Vogelzangs N, van Hout HP, et al. Depressive and anxiety disorders and risk of subclinical atherosclerosis Findings from the Netherlands Study of Depression and Anxiety (NESDA). J Psychosom Res 2010;69(2):203-10. doi: 10.1016/j.jpsychores.2010.01.005 [published Online First: 2010/07/14]

13. Camacho A, McClelland RL, Delaney JA, et al. Antidepressant Use and Subclinical Measures of Atherosclerosis: The Multi-Ethnic Study of Atherosclerosis. J Clin Psychopharmacol 2016;36(4):340-6. doi: 10.1097/JCP.0000000000000518 [published Online First: 2016/06/09]

14. Valeri L, Vanderweele TJ. Mediation analysis allowing for exposure-mediator interactions and causal interpretation: theoretical assumptions and implementation with SAS and SPSS macros. Psychol Methods 2013;18(2):137-50. doi: 10.1037/a0031034 [published Online First: 2013/02/06]

Submitted filename: ResponseLetter_PONE_D_19_27607.docx

Click here for additional data file.

6 Mar 2020

Potential gains in life expectancy from reducing amenable mortality among people diagnosed with serious mental illness in the United Kingdom

PONE-D-19-27607R1

Dear Dr. Dregan,

We are pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it complies with all outstanding technical requirements.

Within one week, you will receive an e-mail containing information on the amendments required prior to publication. When all required modifications have been addressed, you will receive a formal acceptance letter and your manuscript will proceed to our production department and be scheduled for publication.

Shortly after the formal acceptance letter is sent, an invoice for payment will follow. To ensure an efficient production and billing process, please log into Editorial Manager at https://www.editorialmanager.com/pone/, click the "Update My Information" link at the top of the page, and update your user information. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, you must inform our press team as soon as possible and no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

With kind regards,

Sinan Guloksuz, M.D., Ph.D.

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

11 Mar 2020

PONE-D-19-27607R1

Potential gains in life expectancy from reducing amenable mortality among people diagnosed with serious mental illness in the United Kingdom

Dear Dr. Dregan:

I am pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

For any other questions or concerns, please email plosone@plos.org.

Thank you for submitting your work to PLOS ONE.

With kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Sinan Guloksuz

Academic Editor

PLOS ONE