Comparing statistical methods for analyzing skewed longitudinal count data with many zeros: an example of smoking cessation

An Introduction and Practical Guide to Strategies for Analyzing Longitudinal Data in Clinical Trials of Smoking Cessation Treatment: Beyond Dichotomous Point-Prevalence Outcomes

Conceptualizing tobacco dependence as a chronic relapsing condition suggests the need to use analytic strategies that reflect that premise. However, clinical trials for smoking cessation typically define the primary endpoint as a measure of abstinence at a single timepoint distal to the intervention, typically 3-12 months. This reinforces the concept of tobacco outcomes as a dichotomous state-one is, or is not, abstinent. Fortunately, there are several approaches available to handle longitudinal data that reflect the relapsing and remitting nature of tobacco use during treatment studies. In this paper, sponsored by the Society for Research on Nicotine and Tobacco's Treatment Research Network, we present an introductory overview of these techniques and their application in smoking cessation clinical trials. Topics discussed include models to examine abstinence outcomes (eg, trajectory models of abstinence, models for transitions in smoking behavior, models for time to event), models that examine reductions in tobacco use, and models to examine joint outcomes (eg, examining changes in the use of more than one tobacco product). Finally, we discuss three additional relevant topics (ie, heterogeneity of effects, handling missing data, and power and sample size) and provide summary information about the type of model that can be used based on the type of data collected and the focus of the study. We encourage investigators to familiarize themselves with these techniques and use them in the analysis of data from clinical trials of smoking cessation treatment. Implications Clinical trials of tobacco dependence treatment typically measure abstinence 3-12 months after participant enrollment. However, because smoking is a chronic relapsing condition, these measures of intervention success may not accurately reflect the common trajectories of tobacco abstinence and relapse. Several analytical techniques facilitate this type of outcome modeling. This paper is meant to be an introduction to these concepts and techniques to the global nicotine and tobacco research community including which techniques can be used for different research questions with visual summaries of which types of models can be used for different types of data and research questions.

The role of IQOS risk perceptions on cigarette smoking behaviours: results from a prospective pilot study

BACKGROUND

IQOS is a heated tobacco product marketed as an alternative to combustible cigarette smoking. Little is known about cigarette smokers' IQOS health risk perceptions and if these risk perceptions impact IQOS use and cigarette smoking behaviour.

METHODS

Adult, daily, non-treatment-seeking cigarette smokers (n=27), naïve to IQOS, were recruited from Philadelphia, Pennsylvania, USA. Participants were introduced to IQOS and then completed measures of risk perceptions. Participants were given an IQOS 2.4 holder, charger and HeatSticks, and asked to switch completely from cigarettes to IQOS for 14 days. The effects of risk perceptions on changes in IQOS use, cigarettes per day (CPD), the substitution of IQOS for cigarettes and motivation to quit smoking were evaluated.

RESULTS

Over the 14-day switch period, CPD significantly decreased (B=-0.18, 95% CI=-0.26 to -0.09, p<0.0001), IQOS use significantly increased (B=0.02, 95% CI=0.00 to 0.03, p=0.042), as did the percentage IQOS HeatSticks that replaced CPD (B=0.02, 95% CI=0.01 to 0.03, p=0.005). Participants who perceived IQOS as less risky than cigarettes used fewer IQOS HeatSticks per day. A lower percentage of IQOS HeatStick substitution for cigarettes was observed for participants with higher versus lower risk perceptions (B=-0.14, 95% CI=-0.28 to -0.01, p=0.042). Motivation to quit increased from a mean of 5.53 to 6.79 on the contemplation ladder from baseline to day 14 (B=1.26, 95% CI=0.54 to 1.97, p<0.001).

CONCLUSIONS

Smokers reduced their smoking rate and increased motivation to quit smoking while using IQOS. IQOS risk perceptions did not directly account for reductions in smoking, although they may contribute indirectly through increased IQOS use.

Association of clozapine treatment and rate of methamphetamine or amphetamine relapses and abstinence among individuals with concurrent schizophrenia spectrum and amphetamine use disorder: A retrospective cohort study

BACKGROUND

Preliminary evidence suggest clozapine is associated with more favorable impact on concurrent substance use disorder related outcomes in patients with concurrent schizophrenia spectrum disorders (SSD). At the same time, there is a dearth of evidence with regards to clozapine outcomes in the context of concurrent methamphetamine or amphetamine use disorder (MAUD).

AIMS

To examine whether clozapine use decreases rate of methamphetamine or amphetamine (MA) relapses and increases the likelihood of maintaining abstinence from any MA use.

METHODS

A descriptive-analytic retrospective cohort study was conducted on individuals with SSD-MAUD in an inpatient provincial treatment and rehabilitation center for concurrent disorders. Antipsychotic exposure was categorized as "on clozapine" or "on other antipsychotic(s)." Data were collected using electronic health records. Logistic regression was used to examine association of clozapine treatment with likelihood of complete abstinence from MA use for the duration of antipsychotic exposure. Negative binomial regression was used to examine association of clozapine treatment with rate of MA relapses for the duration of antipsychotic exposure.

RESULTS

The majority of the 87 included patients were male. Ethnicity was diverse, with the largest groups self-identifying as Indigenous and European. Clozapine use was both associated with increased likelihood of maintaining abstinence from MA use (adjusted odds ratio (aOR) = 3.05, 95% confidence intervals (CI) = 1.15-8.1, p = 0.025), and decreased rate of MA relapses (aRR = 0.45, 95% CI = 0.25-0.82, p = 0.009) for the duration of antipsychotic exposure. Co-prescription of psychostimulants was associated with increased rate of MA relapses (aRR = 2.43, 95% CI = 1.16-5.10, p = 0.019).

CONCLUSION(S)

In this study, clozapine use compared with other antipsychotics in SSD was associated with improved outcomes related to severe concurrent MAUD. Co-prescription of psychostimulant medications was associated with a poor outcome.

Dealing with Zeros: Adolescent Drug Use, Perceived Disapproval, and Perceived Harm

Because studies examining youth drug use often have data with a high proportion of zeros, they often do not meet the assumptions for univariate or linear regression analyses that are typically used. We demonstrate the use of zero-inflated negative binomial regression models to address excessive zeros in drug use frequency on perceptions of disapproval and perceived harm among middle and high school students (N = 522). We found that perceptions of parent disapproval were a better predictor of marijuana use (p = .01) than peer disapproval. Perceived harm was related to marijuana use (p = .04). Researchers should consider using zero-inflated negative binomial regression models when examining youth drug use.

Promoting appropriate midline catheter and PICC placement through implementation of an EHR-based clinical decision support tool: An interrupted time-series analysis

BACKGROUND

Inappropriate use of peripherally inserted central catheters (PICCs), including multilumen PICCs, may increase the risk of patient complications.

OBJECTIVE

Implement computerized decision support to: (a) increase the use of midline catheters over PICCs when appropriate and (b) decrease PICC lumens when a PICC is necessary.

DESIGNS

Quasi-experimental, interrupted time series.

SETTING

Single tertiary academic medical center.

PATIENTS

Hospitalized adults who received a midline or PICC during clinical care.

INTERVENTION

Decision support on appropriate vascular access device choice via a set of electronic orders embedded within the electronic health record.

MEASUREMENTS

Proportion of midline catheter and single-lumen PICCs placed between January 2, 2017 and November 19, 2017 (preintervention) versus November 20, 2017 to December 1, 2019 (postintervention).

RESULTS

A total of 8758 midline and PICCs were inserted during the study period. A statistically significant increase in the insertion of midline catheters was observed during the intervention (p = .006). In parallel, single-lumen PICC insertion as a proportion of all PICCs also increased after the intervention (p = .035). Results were consistent across multiple disciplines, including internal medicine, surgery, and intensive care. After considering the hospital census, total PICC and midline utilization rates and rates of complications did not change over time. However, an increase in catheter exchanges from less to more invasive devices occurred.

CONCLUSIONS

In this single-center study, the implementation of electronic decision support to inform appropriate catheter use led to a more appropriate midline catheter and single-lumen PICC insertion. Dissemination of this single, effective intervention to examine efficacy in other hospital settings would be welcomed.

Finite mixture Negative Binomial-Lindley for modeling heterogeneous crash data with many zero observations

Crash data are often highly dispersed; it may also include a large amount of zero observations or have a long tail. The traditional Negative Binomial (NB) model cannot model these data properly. To overcome this issue, the Negative Binomial-Lindley (NB-L) model has been proposed as an alternative to the NB to analyze data with these characteristics. Research studies have shown that the NB-L model provides a superior performance compared to the NB when data include numerous zero observations or have a long tail. In addition, crash data are often collected from sites with different spatial or temporal characteristics. Therefore, it is not unusual to assume that crash data are drawn from multiple subpopulations. Finite mixture models are powerful tools that can be used to account for underlying subpopulations and capture the population heterogeneity. This research documents the derivations and characteristics of the Finite mixture NB-L model (FMNB-L) to analyze data generated from heterogeneous subpopulations with many zero observations and a long tail. We demonstrated the application of the model to identify subpopulations with a simulation study. We then used the FMNB-L model to estimate statistical models for Texas four-lane freeway crashes. These data have unique characteristics; it is highly dispersed, have many locations with very large number of crashes, as well as significant number of locations with zero crash. We used multiple goodness-of-fit metrics to compare the FMNB-L model with the NB, NB-L, and the finite mixture NB models. The FMNB-L identified two subpopulations in datasets. The results show a significantly better fit by the FMNB-L compared to other analyzed models.

Within-person longitudinal associations between electronic nicotine delivery systems use and smoking cessation efforts among U.S. continuing adult cigarette smokers

OBJECTIVES

Prior studies that examined the role of electronic nicotine delivery systems (ENDS) use in smoking cessation have shown divergent conclusions. This study examined the time-course of ENDS-associated smoking abstinent behaviors among continuing cigarette smokers who were willing but unable to quit smoking.

METHODS

Data were drawn from the four waves of the Population Assessment of Tobacco and Health Study. Of the 1,684 smokers who tried to quit smoking completely at Wave 1, a total of 1,094 who were continuing smokers until Wave 4 and were not lost to follow-ups comprised the sample. Using generalized linear mixed modeling, we fitted weighted negative binomial regression models to examine within-person associations of ENDS use with quit attempts and number of days abstinent from smoking.

RESULTS

Quit attempt frequency and smoking abstinent days were highest at Wave 1, dropped at Wave 2, and then either increased a little or remained stagnant in later waves. ENDS use to quit smoking was associated with more frequent quit attempts (aIRR=1.67, 95% CI=1.42-1.95) and more abstinent days (aIRR=3.28, 95% CI=2.43-4.44), and the magnitude of such associations became stronger over time. ENDS use was associated with becoming a non-daily smoker among baseline daily smokers.

CONCLUSIONS

ENDS use among continuing smokers may help increase the number of quit attempts and smoking abstinent days. Given that the study sample is continuing smokers who failed in complete smoking cessation, future research would be desirable that evaluates whether such abstinent behaviors sustain and manifest harm reduction with improved health outcomes.

IMPLICATIONS

Smokers increasingly adopt electronic nicotine delivery systems (ENDS) to quit smoking. However, the association of ENDS use with smoking cessation efforts among continuing cigarette smokers who were willing but unable to quit smoking is largely unknown. This study found that quit attempts and smoking abstinent days increased with ENDS use to quit smoking by following up with population-representative continuing smokers. Additionally, this study assessed whether cigarette smokers' abstinent behaviors vary with their use of ENDS by estimating within-person associations with frequent assessments of both exposures and outcomes in a long-term perspective.

A Primary Care Telehealth Pilot Program to Improve Access: Associations with Patients' Health Care Utilization and Costs

Background: The Veterans Health Administration (VHA) piloted an innovative video telehealth program called Virtual Integrated Multisite Patient Aligned Care Teams (V-IMPACT) in fiscal year (FY) 2014. V-IMPACT set up one regional "hub" site where primary care (PC) teams provided regular PC through telehealth services to patients in outlying "spoke" sites that experienced gaps in provider coverage. We evaluated associations between clinic-level adoption of V-IMPACT and patients' utilization and VHA's costs for primary, emergency, and inpatient care. Materials and Methods: This observational study used repeated cross-sections of 208,612 unique veteran patients assigned to a PC team in 22 V-IMPACT spoke sites from FY2013 to FY2018. V-IMPACT adoption in a spoke site was indicated if more than 1% of patients assigned to PC in a site used V-IMPACT services during the year. Association between V-IMPACT adoption and outcomes were assessed using mixed-effects models. Results: V-IMPACT adoption was associated with increased telehealth visits for PC (incidence rate ratio [IRR] = 2.42 [1.29 to 4.55]) and for primary care mental health integration (IRR = 7.25 [2.69 to 19.54]). V-IMPACT adoption was not associated with in-person visits, or with total visits (in-person plus video telehealth). V-IMPACT adoption was also not associated with acute hospital stays, emergency department visits, or VHA costs. Conclusions: Programs such as VHA's V-IMPACT can increase telehealth visits for PC, allowing successful transition across modalities and facilitating continuity of care without impacting total care. Programs should track substitution of in-person visits with telehealth visits and examine its effects on patients' health outcomes, satisfaction, and travel costs.

Effectiveness of a Dyadic Buddy App for Smoking Cessation: Randomized Controlled Trial

BACKGROUND

Tobacco smoking is one of the biggest public health threats. Smartphone apps offer new promising opportunities for supporting smoking cessation in real time. This randomized controlled trial investigated the effectiveness of an app that encourages individuals to quit smoking with the help of a social network member (buddy) in daily life.

OBJECTIVE

The objective of this study is to test the effectiveness of the SmokeFree buddy app compared with a control group with self-reported smoking abstinence and carbon monoxide (CO)-verified smoking abstinence as primary outcomes and self-reports of smoked cigarettes per day (CPD) as a secondary outcome.

METHODS

A total of 162 adults who smoked participated in this single-blind, two-arm, parallel-group, intensive longitudinal randomized controlled trial. Around a self-set quit date (ie, 7 days before the self-set quit date and 20 days after) and 6 months later, participants of the intervention and control groups reported on daily smoking abstinence and CPD in end-of-day diaries. Daily smoking abstinence was verified via daily exhaled CO assessments. This assessment was administered via an app displaying results of exhaled CO, thus addressing self-monitoring in both groups. In addition, participants in the intervention group used the SmokeFree buddy app, a multicomponent app that facilitates social support from a buddy of choice.

RESULTS

A significant reduction in CPD from baseline to the 6-month follow-up was observed among participants in both groups. Multilevel analyses revealed no significant intervention effect on self-reported and CO-verified daily smoking abstinence at the quit date and 3 weeks later. However, CPD was lower at the quit date and 3 weeks later in the intervention group than in the control group. No significant differences between groups were found for any outcome measures 6 months after the quit date. Overall, low app engagement and low perceived usefulness were observed.

CONCLUSIONS

Despite some encouraging short-term findings on the amount of smoking, the SmokeFree buddy app did not have beneficial effects on smoking abstinence over and above the self-monitoring control condition. Future studies should examine whether and what support processes can be effectively stimulated and how app use can be improved to better achieve this goal.

TRIAL REGISTRATION

ISRCTN Registry 11154315; https://www.isrctn.com/ISRCTN11154315.

INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID)

RR2-10.1186/s12889-019-7723-z.

Daily support seeking as coping strategy in dual-smoker couples attempting to quit

OBJECTIVE

Smoking cessation is a stressful event and lapses are frequent. The dynamic model of relapse has been criticized for not integrating interpersonal factors as phasic influences. Seeking social support, as a coping strategy to deal with cravings, may help to refrain from smoking.

DESIGN

Overall, 83 heterosexual dual-smoker couples reported on their craving, the extent of seeking social support from one's partner regarding smoking cessation, and their number of cigarettes smoked in smartphone-based end-of-day diaries, from a joint self-set quit date on across 22 consecutive days.

MAIN OUTCOME MEASURE

Number of cigarettes smoked.

RESULTS

Multilevel analyses indicated that on days with higher-than-average levels of craving, male and female smokers reported more cigarettes smoked. Higher-than-usual support seeking was related to fewer cigarettes smoked that same day. For women only, we found a within-person interaction between craving and support seeking on smoking. On days with higher-than-average support-seeking, the effect of craving on smoking was attenuated.

CONCLUSION

Findings confirm the relevance of interpersonal processes in the relapse process, such as support seeking as coping behavior. Further, as a 'first act' in initiating supportive interactions, support seeking is an important piece in the social support process and a promising target for interventions.

Too many zeros and/or highly skewed? A tutorial on modelling health behaviour as count data with Poisson and negative binomial regression

Background: Dependent variables in health psychology are often counts, for example, of a behaviour or number of engagements with an intervention. These counts can be very strongly skewed, and/or contain large numbers of zeros as well as extreme outliers. For example, 'How many cigarettes do you smoke on an average day?' The modal answer may be zero but may range from 0 to 40+. The same can be true for minutes of moderate-to-vigorous physical activity. For some people, this may be near zero, but take on extreme values for someone training for a marathon. Typical analytical strategies for this data involve explicit (or implied) transformations (smoker v. non-smoker, log transformations). However, these data types are 'counts' (i.e. non-negative whole numbers) or quasi-counts (time is ratio but discrete minutes of activity could be analysed as a count), and can be modelled using count distributions - including the Poisson and negative binomial distribution (and their zero-inflated and hurdle extensions, which alloweven more zeros). Methods: In this tutorial paper I demonstrate (in R, Jamovi, and SPSS) the easy application of these models to health psychology data, and their advantages over alternative ways of analysing this type of data using two datasets - one highly dispersed dependent variable (number of views on YouTube, and another with a large number of zeros (number of days on which symptoms were reported over a month). Results: The negative binomial distribution had the best fit for the overdispersed number of views on YouTube. Negative binomial, and zero-inflated negative binomial were both good fits for the symptom data with over-abundant zeros. Conclusions: In both cases, count distributions provided not just a better fit but would lead to different conclusions compared to the poorly fitting traditional regression/linear models.

Modelling count data with an excess of zero values applied to childhood bone tumour incidence in Iraq

Bone tumours are rarely found in children and adolescents (0- 19 years old), but there are reports from some provinces in Iraq indicating possible increases in the incidence of childhood bone cancer. Since counts are very low and often zero, or near zero, we fitted zero-inflated Poisson, zero-inflated negative binomial, Poisson hurdle, and negative binomial hurdle regression models to investigate these changes. We used data covering the 2000-2015 period taking age, gender and province into account with the aim of identifying potential health disparities. The results indicate that the zero-inflated Poisson is the most appropriate approach. We also found that, the incidence rate ratio of bone tumours for age groups of 5-9, 10-14 and 15-19 years were 134%, 490% and 723% higher, respectively, compared to the 0-4 year olds. The incidence rate was higher by 49% higher in males compared to females. Compared to 2000-2004, the rate was higher during 2005-2009 and 2010-2015 by 23% and 50%, respectively. In addition, the provinces Al-Muthana and Al-Diwaniyah in the South were found to have a higher incidence rate than other provinces. Join point analysis showed that the age-adjusted incidence rate had a significant, increasing trend, with an average percentage change of 3.1% during 2000-2015. The study suggests that further research into childhood tumours, bone tumours in particular, is needed. Reference to the effect of environmental factors in this group of medical disorders would be of special interest.

How Do People Experience and Respond to Social Control From Their Partner? Three Daily Diary Studies

Positive and negative forms of social control are commonly used to regulate another person’s health-related behaviors, especially in couples. Social control efforts have been shown to result in desirable, but also undesirable effects on different outcomes. Little is known for which outcomes, when, and under which contextual conditions these different effects unfold in people’s everyday lives. Using the dual-effects model of health-related social control, we predicted that same-day and previous-day positive social control would result in desirable effects on target behavior, and same-day positive control on affect. Same-day and previous-day negative control was assumed to result in undesirable effects on reactant responses (i.e., doing the opposite of what the partner wanted and hiding the unhealthy behavior), and same-day negative control on affect. Further, we explored whether it makes a difference if one or both partners intend to change their health behavior. Three daily diary studies addressed these questions for smoking (Studies 1 and 2), and physical activity (Study 3). Receiving more positive control related to more desirable target behavior, and feeling better; more negative control was associated with more reactant responses and feeling worse. Social control unfolded its effects within 1 day, but hardly across days, indicating that control and its reactions to it are fast-acting processes in daily life. The pattern of results were the same for couples with one and both partners intending to change their behavior. Further, results replicated when using partner-reported provided control. Based on these results, social control cannot be unanimously recommended as a behavior change strategy in couples. Future studies should follow up on dyadic and temporal dynamics of social control in couples’ everyday lives in different contexts.

Size-dependent male mating tactics and their morphological correlates in Poecilia gillii

Male alternative reproductive strategies are found in some species of most major animal taxa but are especially widespread in fishes. Mature males of the shortfin molly, Poecilia gillii, display extensive variation in size and morphology. We devised a field test of a priori hypotheses regarding the interrelationships between male size, coloration, morphology and mating tactics. Males did not occur in discrete size classes, but instead occurred in a size and morphological continuum. Large males exhibited darker and more orange-coloured dorsal and caudal fins, whereas small males exhibited lighter and more inconspicuous fin coloration. Furthermore, larger males had proportionately deeper bodies, larger dorsal and caudal fins and shorter gonopodia than smaller males. Our field study of male mating behaviour revealed a lack of courtship in this species, and similar levels of mating attempts (gonopodial thrusts) irrespective of male size. Instead, small males were significantly more likely to chase females than were large males. In contrast, large males exhibited higher rates of gonoporal nibbling (a likely means by which males determine, through chemical factors, whether a female is carrying fertilizable ova) and higher likelihood of chasing other males away. In total, we found evidence for the predicted associations between male size, coloration, morphology and mating behaviour. These associations appear likely to maximize mating success for males of a given body size and phenotype.

Smoking cessation with smartphone applications (SWAPP): study protocol for a randomized controlled trial

Background

Tobacco smoking remains one of the biggest public health threats. Smartphone apps offer new promising opportunities for supporting smoking cessation in real-time. The social context of smokers has, however, been neglected in smartphone apps promoting smoking cessation. This randomized controlled trial investigates the effectiveness of a smartphone app in which smokers quit smoking with the help of a social network member.

Methods

This protocol describes the design of a single-blind, two-arm, parallel-group, intensive longitudinal randomized controlled trial. Participants of this study are adult smokers who smoke at least one cigarette per day and intend to quit smoking at a self-set quit date. Blocking as means of group-balanced randomization is used to allocate participants to intervention or control conditions. Both intervention and control group use a smartphone-compatible device for measuring their daily smoking behavior objectively via exhaled carbon monoxide. In addition, the intervention group is instructed to use the SmokeFree Buddy app, a multicomponent app that also facilitates smoking-cessation specific social support from a buddy over a smartphone application. All participants fill out a baseline diary for three consecutive days and are invited to the lab for a background assessment. They subsequently participate in an end-of-day diary phase from 7 days before and until 20 days after a self-set quit date. Six months after the self-set quit date a follow-up diary for three consecutive days takes place. The primary outcome measures are daily self-reported and objectively-assessed smoking abstinence and secondary outcome measures are daily self-reported number of cigarettes smoked.

Discussion

This is the first study examining the effectiveness of a smoking cessation mobile intervention using the SmokeFree Buddy app compared to a control group in a real-life setting around a self-set quit date using a portable objective measure to assess smoking abstinence. Opportunities and challenges with running studies with smoking participants and certain design-related decisions are discussed.

Trial registration

This trial was prospectively registered on 04/04/2018 at ISRCTNregistry: ISRCTN11154315.

Statistical models for longitudinal zero-inflated count data: application to seizure attacks

Background

Chronic non-communicable diseases:- such as epilepsy, are increasingly recognized as public health problems in developing and African countries. This study aimed at finding determinants of the number of epileptic seizure attacks using different count data modeling techniques.

Methods

Four common fixed-effects Poisson family models were reviewed to analyze the count data with a high proportion of zeros in longitudinal outcome, i.e., the number of seizure attacks in epilepsy patients. This is because, in addition to the problem of extra zeros, the correlation between measurements upon the same patient at different occasions needs to be taken into consideration.

Results

The investigation remarkably identified some important factors associated with epileptic seizure attacks. As people grow old, the number of seizure attacks increased and male patients had more seizures than their female counterparts. In general, a patient's age, sex, monthly income, family history of epilepsy andservice satisfaction were some of the significant factors responsible for the frequency of seizure attacks (P value<0.05).

Conclusion

This study suggests that zero-inflated negative binomial is the best model for predicting and describing the number of seizure attacks as well as identifying the potential risk factors. Addressing these risk factors will definitely contain the progression of seizure attack.

Models for analyzing zero-inflated and overdispersed count data: an application to cigarette and marijuana use

INTRODUCTION

This paper describes different methods for analyzing counts and illustrates their use on cigarette and marijuana smoking data.

METHODS

The Poisson, zero-inflated Poisson (ZIP), hurdle Poisson (HUP), negative binomial (NB), zero-inflated negative binomial (ZINB) and hurdle negative binomial (HUNB) regression models are considered. The different approaches are evaluated in terms of the ability to take into account zero-inflation (extra zeroes) and overdispersion (variance larger than expected) in count outcomes, with emphasis placed on model fit, interpretation, and choosing an appropriate model given the nature of the data. The illustrative data example focuses on cigarette and marijuana smoking reports from a study on smoking habits among youth e-cigarette users with gender, age, and e-cigarette use included as predictors.

RESULTS

Of the 69 subjects available for analysis, 36% and 64% reported smoking no cigarettes and no marijuana, respectively, suggesting both outcomes might be zero-inflated. Both outcomes were also overdispersed with large positive skew. The ZINB and HUNB models fit the cigarette counts best. According to goodness-of-fit statistics, the NB, HUNB, and ZINB models fit the marijuana data well, but the ZINB provided better interpretation.

CONCLUSION

In the absence of zero-inflation, the NB model fits smoking data well, which is typically overdispersed. In the presence of zero-inflation, the ZINB or HUNB model is recommended to account for additional heterogeneity. In addition to model fit and interpretability, choosing between a zero-inflated or hurdle model should ultimately depend on the assumptions regarding the zeros, study design, and the research question being asked.

IMPLICATIONS

Count outcomes are frequent in tobacco research and often have many zeros and exhibit large variance and skew. Analyzing such data based on methods requiring a normally distributed outcome are inappropriate and will likely produce spurious results. This study compares and contrasts appropriate methods for analyzing count data, specifically those with an over-abundance of zeros, and illustrates their use on cigarette and marijuana smoking data. Recommendations are provided.

Analysing risk factors of co-occurrence of schistosomiasis haematobium and hookworm using bivariate regression models: Case study of Chikwawa, Malawi

Schistosomiasis and soil-transmitted helminth (STH) infections constitute a major public health problem in many parts of sub-Saharan Africa. In areas where prevalence of geo-helminths and schistosomes is high, co-infection with multiple parasite species is common, resulting in disproportionately elevated burden compared with single infections. Determining risk factors of co-infection intensity is important for better design of targeted interventions. In this paper, we examined risk factors of hookworm and S. haematobium co-infection intensity, in Chikwawa district, southern Malawi in 2005, using bivariate count models. Results show that hookworm and S. haematobium infections were much localised with small proportion of individuals harbouring more parasites especially among school-aged children. The risk of co-intensity with both hookworm and S. haematobium was high for all ages, although this diminished with increasing age, increased with fishing (hookworm: coefficient. = 12.29; 95% CI = 11.50-13.09; S. haematobium: 0.040; 95% CI = 0.0037, 3.832). Both infections were abundant in those with primary education (hookworm: coef. = 0.072; 95% CI = 0.056, 0.401 and S. haematobium: coef. = 0.286; 95% CI = 0.034, 0.538). However, much lower risk was observed for those who were farmers (hookworm: coef. = - 0.349, 95% CI = - 0.547,-0.150; S. haematobium: coef. - 0.239, 95% CI = - 0.406, - 0.072). In conclusion, our findings suggest that efforts to control helminths infection should be co-integrated and health promotion campaigns should be aimed at school-going children and adults who are in constant contact with water.

Measurement error and outcome distributions: Methodological issues in regression analyses of behavioral coding data

Behavioral coding is increasingly used for studying mechanisms of change in psychosocial treatments for substance use disorders (SUDs). However, behavioral coding data typically include features that can be problematic in regression analyses, including measurement error in independent variables, non normal distributions of count outcome variables, and conflation of predictor and outcome variables with third variables, such as session length. Methodological research in econometrics has shown that these issues can lead to biased parameter estimates, inaccurate standard errors, and increased Type I and Type II error rates, yet these statistical issues are not widely known within SUD treatment research, or more generally, within psychotherapy coding research. Using minimally technical language intended for a broad audience of SUD treatment researchers, the present paper illustrates the nature in which these data issues are problematic. We draw on real-world data and simulation-based examples to illustrate how these data features can bias estimation of parameters and interpretation of models. A weighted negative binomial regression is introduced as an alternative to ordinary linear regression that appropriately addresses the data characteristics common to SUD treatment behavioral coding data. We conclude by demonstrating how to use and interpret these models with data from a study of motivational interviewing. SPSS and R syntax for weighted negative binomial regression models is included in online supplemental materials.

Daily negative affect and smoking after a self-set quit attempt: The role of dyadic invisible social support in a daily diary study

OBJECTIVES

Social support receipt from one's partner is assumed to be beneficial for successful smoking cessation. However, support receipt can have costs. Recent research suggests that the most effective support is unnoticed by the receiver (i.e., invisible). Therefore, this study examined the association between everyday levels of dyadic invisible emotional and instrumental support, daily negative affect, and daily smoking after a self-set quit attempt in smoker-non-smoker couples.

METHODS

Overall, 100 smokers (72.0% men, mean age M = 40.48, SD = 9.82) and their non-smoking partners completed electronic diaries from a self-set quit date on for 22 consecutive days, reporting daily invisible emotional and instrumental social support, daily negative affect, and daily smoking.

RESULTS

Same-day multilevel analyses showed that at the between-person level, higher individual mean levels of invisible emotional and instrumental support were associated with less daily negative affect. In contrast to our assumption, more receipt of invisible emotional and instrumental support was related to more daily cigarettes smoked.

CONCLUSIONS

The findings are in line with previous results, indicating invisible support to have beneficial relations with affect. However, results emphasize the need for further prospective daily diary approaches for understanding the dynamics of invisible support on smoking cessation. Statement of contribution What is already known on this subject? Social support receipt from a close other has proven to have emotional costs. According to current studies, the most effective social support is unnoticed by the receiver (i.e., invisible). There is empirical evidence for beneficial effects of invisible social support on affective well-being. What does this study add? Confirming benefits of invisible social support for negative affect in a health behaviour change setting Providing first evidence for detrimental effects of invisible social support on smoking.

Analyzing count variables in individuals and groups: Single level and multilevel models

We address two challenges in data analysis of group research. First, data may be clustered (i.e., responses of individual group members are correlated). Second, some dependent variables may consist of integer counts of number of occurrences of an event. Familiar ANOVA and regression models provide nonoptimal analyses in both cases. Standard multilevel (mixed) models yield accurate inference for clustered normally distributed data. Generalized linear models (GLMs), specifically Poisson regression and related models, yield accurate inference for nonclustered count data. New generalized linear mixed models (GLMMs) integrate GLMs with multilevel models, addressing both challenges and yielding accurate inferences for grouped count outcomes. To provide the necessary background for understanding GLMMs, we first introduce GLMs, with detailed coverage in an example of Poisson regression. We then introduce multilevel models. Finally, we develop GLMMs and illustrate in an example their application to clustered count data. Group research may benefit from the flexibility provided by GLMMs.

A Dyadic Action Control Trial in Overweight and Obese Couples (DYACTIC)

Background

Enhancing physical activity in overweight and obese individuals is an important means to promote health in this target population. The Health Action Process Approach (HAPA), which was the theoretical framework of this study, focuses on individual self-regulation variables for successful health behavior change. One key self-regulation variable of this model is action control with its three subfacets awareness of intentions, self-monitoring and regulatory effort. The social context of individuals, however, is usually neglected in common health behavior change theories. In order to integrate social influences into the HAPA, this randomized controlled trial investigated the effectiveness of a dyadic conceptualization of action control for promoting physical activity.

Methods/Design

This protocol describes the design of a single-blind randomized controlled trial, which comprises four experimental groups: a dyadic action control group, an individual action control group and two control groups. Participants of this study are overweight or obese, heterosexual adult couples who intend to increase their physical activity. Blocking as means of a gender-balanced randomization is used to allocate couples to conditions and partners to either being the target person of the intervention or to the partner condition. The ecological momentary intervention takes place in the first 14 days after baseline assessment and is followed by another 14 days diary phase without intervention. Follow-ups are one month and six months later. Subsequent to the six-months follow-up another 14 days diary phase takes place.

The main outcome measures are self-reported and accelerometer-assessed physical activity. Secondary outcome measures are Body Mass Index (BMI), aerobic fitness and habitual physical activity.

Discussion

This is the first study examining a dyadic action control intervention in comparison to an individual action control condition and two control groups applying a single-blind randomized control trial. Challenges with running couples studies as well as advantages and disadvantages of certain design-related decisions are discussed. This RCT was funded by the Swiss National Science Foundation (PP00P1_133632/1) and was registered on 27/04/2012 at http://www.isrctn.com/ISRCTN15705531.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2458-14-1321) contains supplementary material, which is available to authorized users.