Lifestyle Coaching May Be an Effective Treatment for Schizophrenia

Abstract

This commentary critiques the Danish CHANGE trial, which evaluated 3 levels of outpatient intervention intensity, in a group of outpatients with obesity and schizophrenia. Neither adding care coordination with weekly nurse contacts alone nor combining this treatment with assertive community lifestyle coaching as compared to treatment as usual improved outcomes, which included cardiovascular disease risk calculation, cardiorespiratory fitness, weight, and self-reported behaviors such as smoking, physical activity, and diet. The CHANGE trial investigators appear strongly averse to recommending the development and implementation of lifestyle medicine programs as a major component when treating outpatients with severe mental disorders. The potential dismissal of lifestyle medicine as a component of treatment for severe mental disorders is problematic. Valuable lessons can be learned from more thoroughly analyzing secondary outcomes such as medical and psychiatric hospitalization rates and total health care cost. The CHANGE trial data analysis needs to be expanded beyond the focus on changes in weight and serum cholesterol. Insulin resistance and high refined carbohydrate intake may be major factors in determining both the medical and psychiatric clinical course of schizophrenia. Assertive community lifestyle coaching is a novel treatment modality. Evidence strongly suggests assertive community lifestyle coaching substantially decreases both psychiatric and medical hospitalization rates.

“Assertive community lifestyle coaching is an important outgrowth of the assertive community treatment team movement.”

The World Health Organization guidelines for managing physical health conditions in adults with severe mental disorders (SMDs) examine how standard psychiatric pharmacological treatments and patient lifestyle are related to the 10-20 year reduction in life expectancy and 2-3 fold higher mortality rates associated with SMD. ¹ The SMD diagnoses, which include major depression, bipolar disorders, schizophrenia, and related spectrum disorders, are treated with medications that can increase appetite, body weight, insulin resistance, and cardio-metabolic risk. Obesity and cardiovascular disease are responsible for the vast majority of these premature deaths. ² Modifiable risky behaviors (tobacco smoking, obesity, physical inactivity, and unhealthy diet) account for much of the early mortality gap associated with SMD. There is recognition of “an urgent need for tailored effective interventions to promote cardiovascular health.” ³ To address this urgent need, we feel more consideration needs to be given to an assertive community lifestyle coaching (ACLC) interventional approach. ACLC is based on a theoretical framework adapted from assertive community treatment, which focuses on utilizing home and community visits rather than office visits to address the many treatment needs of patients with SMD. ⁴ Ironically, the evidence of a promising ACLC intervention may have been undeservedly dismissed by its investigators. ⁵

Our comments address the analyses and reported findings of the Danish CHANGE trial, which tested the impact of assertive lifestyle coaching to reduce the 10-year cardiovascular disease risk score in a group of patients with obesity and schizophrenia. In the CHANGE trial, a total of 428 patients were evenly randomized to one of the following 3 treatment groups for 12 months: The CHANGE group received ACLC plus care coordination with weekly nurse contacts plus treatment as usual. The CARE group received care coordination with weekly nurse contacts plus treatment as usual. The TAU group received only treatment as usual. The study planned to assess the outcomes at 3 time points: baseline, 12 months (post-intervention), and 24 months after randomization. At the end of the study no intervention effects could be found in the calculated risk score of cardiovascular disease, cardiorespiratory fitness, weight, and self-reported behaviors such as smoking, physical activity, and diet. Based on these findings, the investigators “suggested that future research should focus on environmental/structural changes rather than individually anchored interventions.” ⁵ Although the need to “focus on environmental/structural changes” in future research is warranted, the conclusions discourage research investigating “individually anchored interventions” as we enter the era of precision and personalized nutrition research. The findings from other trials conducted on patients with SMD demonstrate that outpatient lifestyle habits can be altered and produce better health outcomes.^6–10 Furthermore, some analytical aspects demand a closer look. Non-adherence is a big issue in any clinical trial, and it is even more so with behavioral interventions. The authors of the CHANGE trial reported that only 60% of the patients assigned to CHANGE intervention had at least half of the intended community or home visits, which could be a significant reason for a diluted treatment effect and no observed treatment benefit. Under this scenario, an as-treated (AT) analysis that works on the principle of actual treatment received combined with the amount of adherence (dose) could have provided a more pragmatic and meaningful measure of treatment differences.^11,12

A 2-year follow-up of the CHANGE trial compared it to the landmark Look AHEAD (Action for Health in Diabetes) study. In this article, the CHANGE trial researchers implied that both studies demonstrate lifestyle health coaching is of little clinical or economic value. ¹³ We believe this does an unwarranted disservice to the findings of both studies.

The Look AHEAD trial was a randomized controlled study comparing Intensive Lifestyle Intervention (ILI) to Diabetes Support and Education (DSE) in overweight and obese type 2 diabetics to track the development of cardiovascular disease over time. The trial intervention was stopped for futility after a median follow-up of 9.6 years. Similar to the CHANGE trial, in the Look AHEAD trial, the ILI intervention did not decrease cardiovascular events. ¹⁴ Indeed, both groups had half the number of projected cardiovascular events, perhaps because both groups had become more cognizant of the value of lifestyle changes. However, there were still many differences between the 2 groups. For example, the ILI group had improved insulin sensitivity, reduced need of diabetes medication, less depression, and better maintenance of physical mobility than the DSE group. ¹⁵ It is of particular interest that the ILI group also had lower hospitalization rates. “Compared with DSE over 10 years, ILI had fewer hospitalizations, fewer medications, and lower health care costs”.^15,16 “ILI led to a reduction in annual hospitalizations (11%, P = .004), hospital days (15%, P = .01), and number of medications (6%, P < .0001) resulting in cost savings for hospitalization (10%, P = .04).” ¹⁶

The medical hospitalization rates in the CHANGE trial were 16.2% for the TAU group, 17.6% for the CARE group, and a substantially lower 12.3% for the CHANGE group. Our unadjusted log-binomial regression model, though not statistically significant, reported a 25% (RR: 075, 95% CI: .43, 1.35, P = .34) and 30% (RR: 0.70, 95% CI: .40, 1.23, P = .22) risk reduction in medical hospitalization among the CHANGE group compared to TAU and CARE groups, respectively. Similarly, the psychiatric hospitalization rates in the CHANGE trial were 24.3% for the TAU group, 33.8% for the CARE group, and a substantially lower 18.8% for the CHANGE group. The unadjusted model suggested a statistically significant risk reduction of 44% (RR: .56, 95% CI: .37, .84, P = .006) in psychiatric hospitalizations in the CHANGE group as compared to the CARE group. A nonsignificant risk reduction of 23% (RR: .77, 95% CI: .49, 1.21) was observed in the CHANGE group compared to the TAU group.

Very importantly, the prevalence of diabetes and high alcohol intake in the CHANGE and CARE groups was not taken into account in the analysis, despite rates that were nearly 2-fold higher in these groups than in the TAU group (diabetes prevalence: CHANGE = 18.6%, CARE = 17.0%, TAU = 9.5%; high alcohol consumption prevalence: CHANGE = 8.0%, CARE = 8.5%, TAU = 4.1%). Both diabetes and high alcohol consumption are associated with an increased risk of medical hospitalization.^17,18 Therefore, it would be reasonable to expect that the CHANGE and CARE groups would have higher medical hospitalization rates than the TAU group. However, the medical hospitalization rates were similar for the CARE and TAU groups (17.6% vs 16.2%) while the CHANGE group had an unexpectedly lower rate of 12.3%. These findings were reported in the text but were not included in the tables that were statistically adjusted for baseline differences. It would have been desirable to have the hospitalization rates also adjusted for diabetes, high alcohol consumption, and other confounders. For example, substance dependence prevalence was greater in the CHANGE group (5.8%) than the TAU group (3.4%) and the CARE group (2.8%).

The effects of therapeutic lifestyle coaching in seriously mentally ill (SMI) patients were also examined in the STRIDE study.^6,7 Green et al ⁷ noted a lower medical hospitalization rate not only in the 12-month intervention period but also in the 12-month follow-up period. The intervention was weekly or monthly group sessions. During the 12-month follow-up period, the medical hospitalization rate was 5.7% for the treatment group vs 21.0% for the control group, P = .004. This occurred even though the patients in the intervention group gained weight and had rising fasting glucose levels during the follow-up period. The one blood parameter that did improve during the follow-up period was the average fasting insulin level, which decreased from 9.343 to 7.912 micromoles per ml.

Green et al stated, “The steady decreases in fasting insulin levels found in this trial suggest that improved insulin sensitivity may help explain the long-term benefits of such lifestyle modification.” Further bolstering this view is research by Sacks et al, ¹⁹ who found the hospitalization rate for insulin-sensitive obese patients was almost 50% lower than the rate for insulin-resistant obese patients. The Look AHEAD trial also found improvement in insulin sensitivity and reduced hospitalization rates in the ILS intervention arm. ¹⁴ We speculate that the increased insulin sensitivity in the STRIDE and ILS treatment groups may be related to improvements in the quality of food intake. The CHANGE trial did not study insulin sensitivity.

There is growing evidence that a high intake of low-quality energy-dense foods, particularly low-quality carbohydrate-dense foods, is associated with decreased insulin sensitivity as well as weight gain.^20,21 Intake of “empty calories” or high-calorie unhealthful food items was evaluated in the SMILES trial, a randomized controlled trial of dietary counseling for adults with major depression. ²² Researchers found a much larger remission rate in the dietary support group than in the control group (32.3% vs 8.0%, P = .028) and noted that “with respect to the consumption of unhealthful food items, intake of extra servings substantially declined (mean decrease 21.76 (SD 16.01) servings/week) in the dietary support group.” ²² All the improvements in the SMILES trial occurred—as in the CHANGE trial—despite no improvement in weight or other conventional biomarkers of health such as serum cholesterol and glucose levels. Although analysis of the CHANGE trial dietary measures found no differences in vegetable and fruit consumption, ⁵ the analysis did not address whether or not there were changes in consumption of empty calories or unhealthful foods.

The 2-year update of the CHANGE trial ¹³ did not provide any more dietary information and only emphasized previous negative conclusions. However, it did incidentally provide further evidence that the assertive lifestyle interventions of the CHANGE trial may have been beneficial. Although the evidence is scanty, the 2-year follow-up data shows notably fewer deaths in the CHANGE group (CHANGE = 2; CARE = 7; TAU = 4) even though the CHANGE group seems to be the least healthy group based on differences in the baseline prevalence of diabetes, high alcohol consumption, and substance dependence. Assuming all participants were followed until 2 years, the unadjusted exact Poisson regression model suggested a mortality rate reduction of 46% (RR: .56, 95% CI: .05, 3.74; P = .75) and 71% (RR: 0.29, 95% CI: .03, 1.54; P = .50) in the CHANGE group compared to the TAU and CARE groups, respectively.

Decreasing hospitalization and mortality rates are a more authentic bottom-line measure of improving health than declining weight or serum cholesterol. The investigators of the CHANGE trial have an opportunity to apply more depth to the analysis of hospitalization rates. This should include psychiatric hospitalization rates, which also received minimal analysis. These unadjusted hospitalization rates already demonstrate at least one statistically significant difference in favor of the CHANGE treatment. Also, in general, instead of hospitalization as a binary response, modeling as a count response (number of hospitalizations or number of days hospitalized) would provide more insights.

The lower psychiatric hospitalization rate found in the CHANGE group ⁵ echoes the findings of a post hoc multiple regression analysis by Malcolm Peet of national food consumption patterns taken from 2 World Health Organization schizophrenia studies.^23–25 Peet concludes, “The most striking relationship in the schizophrenia databases is between sugar consumption and outcome.

This was an independent predictor of poor outcome of schizophrenia as judged by both social and hospital admission criteria and the total outcome. The most consistent finding is that greater consumption of refined sugar is associated with a worse outcome of schizophrenia and a greater prevalence of depression.” ²⁵ Peet also mentions insulin resistance as a possible causal factor and calls for more intervention studies, such as the CHANGE trial, to test the high sugar consumption hypothesis. This hypothesis is compatible with longstanding and recurring literature documenting disturbed carbohydrate metabolism and insulin resistance in antipsychotic-naive patients with schizophrenia.^26–30 Additional analysis of the CHANGE trial data could lead to more information on the impact(s) of the reduction of consumption of empty calories and refined sugar. This information could be retrievable from the Food Frequency Questionnaire and 24-hour recall diet diaries compiled in the study. Decreased consumption of refined sugar in the CHANGE group vs the treatment as usual (TAU) group would support the Peet hypothesis and could have important treatment implications.

As discussed above, the CHANGE trial database seems well suited for further secondary analysis. If the original researchers are not inclined to do it, there is another option. The CHANGE data set is available through Denmark Statistics at Data for research – Statistics Denmark (dst.dk), and grants to restore invisible and abandoned trials (RIAT Grants) have been awarded to independent researchers since 2018. ³¹ RIAT Grant funding is intended—among other purposes—to correct a “distorted representation (distortion), which occurs when publication in medical journals present a biased or misleading description of the design, conduct or results of the trial.” ³² This option might be considered for the CHANGE trial.

Since the publication of the CHANGE trial study in 2016, 3 other distinguished research teams have reported on their attempts to lower the risk of cardiovascular disease in patients with severe mental illness by utilizing lifestyle coaching techniques. These large randomized control trials are the IMPACT, Primrose, and STEPWISE trials.^33–35 None of them applied an assertive community approach. Negative findings predominated with one very relevant exception. In the Primrose trial, when comparing the TAU group with the intervention group, there were fewer psychiatric hospitalizations (statistically significant) and fewer medical hospitalizations (not statistically significant) in the intervention group which led to a statistically significant (P = .012) total health care cost savings over the TAU group. The Primrose investigators considered this “finding worthy of further exploration” and noted that it “might be important.” ³⁴ This most certainly encourages taking another look at the CHANGE trial results.

Assertive community lifestyle coaching is an important outgrowth of the assertive community treatment (ACT) team movement. CHANGE trial subjects receiving ACLC were usually seen in their homes and communities, and rarely in clinics, offices, meeting halls, or gyms. Coaches assertively visited clients. They did not wait for clients to visit them. By adopting this approach, the CHANGE trial is fundamentally different from any of the other lifestyle counseling studies.

When the ACT team concept was developed at the University of Wisconsin in the late 1960s, its designers confronted a common problem in outpatient psychiatry. How can clinicians treat recently discharged patients who do not keep clinic appointments or take prescribed medications or comply with other prescribed treatments? The solution? Engage these patients in their homes and communities. If the patient does not come to the clinic, the clinic will come to the patient. Success was primarily measured by reduced psychiatric re-hospitalization rates, and there was substantial success.^4,36 We believe that the CHANGE trial results demonstrate another example of this success.

A re-analysis of primary and secondary outcomes using an appropriate statistical model and further assessment of hospitalization rates as outcome variables is indicated. If the CHANGE group, after adjustments, had a significant reduction in hospitalization rates, consideration should be given to estimating the potential cost savings of this benefit. If there are savings, this would suggest that an ACT model intervention—such as that used with the CHANGE group—is underutilized and may be clinically and fiscally valuable for some medical-psychiatric cohorts.^37–42