Selection Bias In Estimates Research Articles

Exploring the Effects of Early Censoring and Analysis of Clinical Trial Survival Data on Effectiveness and Cost-effectiveness Estimation through a Case Study in Advanced Breast Cancer.

Interim analyses of clinical trial data are frequently used to provide evidence to obtain marketing authorization for new drugs. However, results from such analyses may not reflect true estimates of relative effectiveness when trial follow-up is complete. Survival results, available at 2 time points from a breast cancer clinical trial, were compared to test the hypothesis that using immature data and a widely used right-censoring rule leads to biased survival estimates. Kaplan-Meier progression-free and overall survival data from 2 published CLEOPATRA trial reports (2012 and 2014) were digitized. Overlaying these results highlighted divergent trends. Parametric functions were fitted to both data sets but did not indicate consistent patterns that could be used as a basis for long-term extrapolation. Heavy censoring of patients in the early data cut coincides with sudden changes in hazard trends and survival patterns, supporting the hypothesis of censoring bias. This challenges the validity of estimates of clinical benefit (progression-free survival and overall survival) based on extrapolation of results from interim analyses of trial data, using a commonly employed censoring rule.

Influence of sex and transients on survival and detection probabilities of the southeastern beach mouse

We used Cormack–Jolly–Seber capture–recapture analyses to investigate differences in monthly survival of the southeastern beach mouse (Peromyscus polionotus niveiventris) between sexes, between juveniles and adults, and effects due to the presence of transients that can bias survival estimates. We accounted for transients (e.g., nonresidents) by distinguishing survival for the initial month after an individuals' 1st capture from monthly survival after the 1st recapture. Survival estimates for 1st captures and recaptures of females were 0.70 and 0.78, respectively. Survival estimates for 1st captures and recaptures of males were 0.74 and 0.80, respectively. Our results showed that a comparably small proportion of transients (females = 0.10; males = 0.08) can bias survival estimates.

Accounting for false mortality in telemetry tag applications

Deaths of animals in the wild are rarely observed directly, which often limits understanding of survival rates. Telemetry transmitters offer field ecologists the opportunity to observe mortality events in cases as the absence of animal movement. When observations of mortality are based on factors such as the absence of animal movement, live individuals can be mistaken for dead, resulting in biased estimates of survival. Additionally, tag failure or emigration might also influence estimates of survival in telemetry studies. Failing to account for mis-classification, tag failure, and emigration rates can result in overestimates of mortality rates by up two-fold, even when the data are corrected for obviously mistaken entries. We use a multi-state capture–recapture model with a misclassification parameter in estimating both the rate of permanent emigration and/or tag failure and the rate at which individuals are mistakenly identified as dead. We use this method on an annual telemetry survey of three species of native fish in the Murray river, Australia: Murray cod (Maccullochella peelii), trout cod (Maccullochella macquariensis) and golden perch (Macquaria ambigua). Evidence for higher mortality rates in the first year post-implantation occurred for Murray cod and golden perch, which is likely an effect of tagging and/or the transmitter, or transmitters shedding. Using simulations, we confirm that our model approach is robust to a broad range of misclassification and transmitter failure rates. With these simulations we also demonstrate that misclassification models that do not account for emigration will likely be erroneous if live and dead animals have different probabilities of detection. These findings will have a broad interest to ecologists wishing to account for multiple sources of misclassification error in capture-mark-recapture studies, with the caveat that the specifics of the approach are dependent on species, transmitter types and other aspects of experimental design which may or may not be amenable to the misclassification framework.

Best Estimate Selection Bias of Estimates of the Value of a Statistical Life

Selection of the best estimates of economic parameters frequently relies on the best estimate or a meta-analysis of the “best-set” of parameter estimates from the literature. Using an all-set dataset consisting of all reported estimates of the value of a statistical life (VSL) as well as a best-set sample of the best estimates from these studies, this article estimates statistically significant publication selection biases in each case. Biases are much greater for the best-set sample, as one might expect, given the subjective nature of the best-set selection process. While “best estimate selection bias” potentially exacerbates the problems associated with existing publication selection biases, the mean and median best-set estimates of the VSL are similar to those in the all-set sample. The bias-corrected estimate of the VSL for the all-set USA sample is $9.5 million and $11.5 million based on the CFOI data.

Competing risks a statistical approach

Survival analyses are commonly applied to study death or other events of interest. In such analyses, so-called competing risks may form an important problem. Competing risk are said to be present when a patient is at risk of more than one mutually exclusive event, such as death from different cause which will prevent any other from happening. when competing events are not independent treating subjects dying from extraneous causes as censored and then computing the ordinary Kaplan-Meir estimate results in biased survival estimates, in such cases,the cumulative incidence and cause-specific hazard may lead to valid results, but the resulting model does not allow one to estimate risks conditional on removal of one or more causes of the event.

Estimating sex specific apparent survival and dispersal of Little Ringed Plovers (Charadrius dubius)

Sex differences in survival have consequences to population dynamics making information on sex specific survival important. In birds, females often have lower survival than males, and one of the main mechanisms is considered to be differential reproductive investment. We studied apparent adult survival and local recruitment of a small monoga-mous shorebird, the Little Ringed Plover (Charadrius dubius), with 18 years of capture– recapture data collected from a population breeding in Southern Finland. We expected no sex differences in adult survival because parental care is shared and double-broods and excessive relaying are rare in this northern population. Because apparent survival is the product of true survival and site fidelity, we also estimated breeding and natal dispersal distances within the study area to examine bias in survival estimates caused by permanent emigration. We found higher apparent adult survival in males (0.660) than in females (0.609), but no sex differences in local recruitment (0.080). Breeding dispersal distances were longer in females than in males mainly due to lower breeding site fidelity of females (0.620) than that of males (0.808). Thus, the lower apparent survival of females likely resulted from permanent emigration. Interestingly, the philopatric portion of the population (from local recruitment analysis) had higher apparent adult survival than what was found from the analysis of all adults. These apparent survival estimates that are less likely to be biased by permanent emigration showed no sex differences (males 0.734; females 0.721), supporting our conclusion of no sex difference in true adult survival.

Population trends and survival of nesting green sea turtles Chelonia mydas on Aves Island, Venezuela

Long-term demographic data are valuable for assessing the effect of anthropogenic impacts on endangered species and evaluating recovery programs. Using a 2-state open robust design model, we analyzed mark-recapture data from green turtles Chelonia mydas sighted be - tween 1979 and 2009 on Aves Island, Venezuela, a rookery heavily impacted by human activities before it was declared a wildlife refuge in 1972. Based on the encounter histories of 7689 nesting females, we estimated the abundance, annual survival, and remigration intervals for this popula- tion. Female survival varied from 0.14−0.91, with a mean of 0.79, which is low compared to sur- vival of other populations from the Caribbean (mean = 0.84) and Australia (mean = 0.95), even though we partially corrected for tag loss, which is known to negatively bias survival estimates. This supports prior suggestions that Caribbean populations in general, and the Aves Island popu- lation in particular, may be more strongly impacted than populations elsewhere. It is likely that nesters from this rookery are extracted while foraging in remote feeding grounds where hunting still occurs. Despite its relatively low survival, the nesting population at Aves Island increased dur- ing the past 30 years from approx. 500 to >1000 nesting females in 2009. Thus, this population, like others in the Caribbean and the Atlantic, seems to be slowly recovering following protective man- agement. Although these findings support the importance of long-term conservation programs aimed at protecting nesting grounds, they also highlight the need to extend management actions to foraging grounds where human activities may still impact green turtle populations.

Evidence for skipped spawning in a potamodromous cyprinid, humpback chub (Gila cypha), with implications for demographic parameter estimates

Our findings reveal evidence for skipped spawning in a potamodromous cyprinid, humpback chub (HBC; Gila cypha). Using closed robust design mark-recapture models, we found, on average, spawning HBC transition to the skipped spawning state (γ¯ˆ″) with a probability of 0.45 (95% CRI (i.e. credible interval): 0.10, 0.80) and skipped spawners remain in the skipped spawning state (γ¯ˆ′) with a probability of 0.60 (95% CRI: 0.26, 0.83), yielding an average spawning cycle of every 2.12 years, conditional on survival. As a result, migratory skipped spawners are unavailable for detection during annual sampling events. If availability is unaccounted for, survival and detection probability estimates will be biased. Therefore, we estimated annual adult survival probability (S), while accounting for skipped spawning, and found S remained reasonably stable throughout the study period, with an average of 0.75 ((95% CRI: 0.66, 0.82), process variance σ2=0.005), while skipped spawning probability was highly dynamic (σ2=0.306). By improving understanding of HBC spawning strategies, conservation decisions can be based on less biased estimates of survival and a more informed population model structure.

Dive to survive: effects of capture depth on barotrauma and post-release survival of Atlantic cod (Gadus morhua) in recreational fisheries

Abstract Atlantic cod (Gadus morhua) caught in recreational fisheries are commonly released, often with barotrauma after rapid decompression. Mouth-hooked, non-bleeding cod kept in a floating net pen showed mortalities ≥40% when angled from >50 m depth, likely because of cumulative stress from ongoing barotrauma and exposure to warm surface water. In a natural setting, however, cod have the opportunity to descend after release and are not restricted to the surface. In a follow-up study, 97.8% of similarly selected cod managed to dive following immediate release, whereas 2.2% were floaters. No mortality was observed for divers kept in cages, which were lowered to capture depth for 72 h. While the floaters would likely have died in a natural setting, no mortality was observed when they were recompressed and kept at capture depth for 72 h. The occurrence of swim bladder ruptures, swollen coelomic cavities, venous gas embolisms, and gas release around the anus was significantly influenced by capture depth (range 0–90 m). A supplementary radiology study showed inflated swim bladders in 87% of the cod after 72 h, and most barotrauma signs had disappeared after 1 month. This study encourages investigation of survival potential for physoclistous species when high mortalities are assumed but undocumented. Matching natural post-release and containment environment is essential in the experimental setup, as failure to do so may bias survival estimates, particularly when a thermocline is present. Assuming minimal predation, short-term mortality of cod experiencing barotrauma is negligible if cod submerge quickly by themselves and are otherwise not substantially injured. Survival of floaters may be increased by forced recompression to capture depth. Sublethal and long-term impacts of barotrauma remain to be studied. To ensure that cod have sufficient energy to submerge, anglers are encouraged to avoid fighting the fish to exhaustion and to minimize handling before release.

Effects of predation on telemetry-based survival estimates: insights from a study on endangered Atlantic salmon smolts

Telemetry is increasingly being used to estimate population-level survival rates. However, these estimates may be affected by the detectability of telemetry tags and are reliant on the assumption that telemetry data represent the movements of the tagged fish. Predation on tagged fish has the potential to bias survival estimates, and unlike the issue of detectability, methods to correct for the resulting bias (termed “predation bias”) are not yet developed. In an acoustic telemetry study on inner Bay of Fundy Atlantic salmon (Salmo salar) smolts during 2008 and 2011, unusual tag detection patterns were indicative that some data may have been representative of the movements of predators rather than smolts. To incorporate predation effects into the resulting survival estimates, a suite of 11 summary migration metrics were compared between Atlantic salmon smolts and striped bass (Morone saxatilis). Cluster analyses revealed that 2.4% to 13.6% of tags implanted in smolts exhibited migration patterns more similar to striped bass than to other smolts, which was interpreted here as evidence of predation. Reassigning the fate of these tags as “depredated–died” reduced estimated survival from 43.5% to 41.1% in 2008 and from 32.6% to 19.0% in 2011 relative to a traditional mark–recapture model, illustrating the effect of predation bias in this case study.

Survival extrapolation in the presence of cause specific hazards.

Health economic evaluations require estimates of expected survival from patients receiving different interventions, often over a lifetime. However, data on the patients of interest are typically only available for a much shorter follow‐up time, from randomised trials or cohorts. Previous work showed how to use general population mortality to improve extrapolations of the short‐term data, assuming a constant additive or multiplicative effect on the hazards for all‐cause mortality for study patients relative to the general population. A more plausible assumption may be a constant effect on the hazard for the specific cause of death targeted by the treatments. To address this problem, we use independent parametric survival models for cause‐specific mortality among the general population. Because causes of death are unobserved for the patients of interest, a polyhazard model is used to express their all‐cause mortality as a sum of latent cause‐specific hazards. Assuming proportional cause‐specific hazards between the general and study populations then allows us to extrapolate mortality of the patients of interest to the long term. A Bayesian framework is used to jointly model all sources of data. By simulation, we show that ignoring cause‐specific hazards leads to biased estimates of mean survival when the proportion of deaths due to the cause of interest changes through time. The methods are applied to an evaluation of implantable cardioverter defibrillators for the prevention of sudden cardiac death among patients with cardiac arrhythmia. After accounting for cause‐specific mortality, substantial differences are seen in estimates of life years gained from implantable cardioverter defibrillators. © 2014 The Authors Statistics in Medicine Published by John Wiley & Sons Ltd.

Comparison of Tributary Survival Estimates of Steelhead using Cormack–Jolly–Seber and Barker Models: Implications for Sampling Efforts and Designs

AbstractWe conducted simulations to compare the precision and bias of survival estimates from Cormack–Jolly–Seber (CJS) and Barker models to known parameter values based on empirical data for steelhead/resident Rainbow Trout Oncorhynchus mykiss from the John Day River, Oregon. We simulated seasonal differences in recapture and survival rates, and we varied the number of fish tagged, recapture and resight rates, sample site size, and fish movement (migratory or resident). Survival estimates from the Barker model had higher precision and lower or equal bias in comparison with estimates from the CJS model under almost all simulation scenarios. The precision of Barker survival estimates increased the most as the number of tagged fish increased from 50 to 200 (CV = 0.4–0.09). The Barker model's superior performance was dependent on the availability of resight data; such data are becoming more readily available, especially in places where large numbers of individuals are PIT‐tagged and where an interrogation infrastructure exists (e.g., Columbia River basin). Tagging of 75–100 fish/site during high‐capture periods (e.g., summer and fall) and focusing on the resighting of fish with fixed or mobile interrogators during low‐capture periods (i.e., winter and spring) may be the most cost‐effective strategy for improving estimates of juvenile steelhead survival.Received April 18, 2014; accepted August 26, 2014

Reducing bias in survival under nonrandom temporary emigration.

Despite intensive monitoring, temporary emigration from the sampling area can induce bias severe enough for managers to discard survival parameter estimates toward the terminus of the times series (terminal bias). Under random temporary emigration, unbiased parameters can be estimated with CJS models. However, unmodeled Markovian temporary emigration causes bias in parameter estimates, and an unobservable state is required to model this type of emigration. The robust design is most flexible when modeling temporary emigration, and partial solutions to mitigate bias have been identified; nonetheless, there are conditions were terminal bias prevails. Long-lived species with high adult survival and highly variable nonrandom temporary emigration present terminal bias in survival estimates, despite being modeled with the robust design and suggested constraints. Because this bias is due to uncertainty about the fate of individuals that are undetected toward the end of the time series, solutions should involve using additional information on survival status or location of these individuals at that time. Using simulation, we evaluated the performance of models that jointly analyze robust design data and an additional source of ancillary data (predictive covariate on temporary emigration, telemetry, dead recovery, or auxiliary resightings) in reducing terminal bias in survival estimates. The auxiliary resighting and predictive covariate models reduced terminal bias the most. Additional telemetry data were effective at reducing terminal bias only when individuals were tracked for a minimum of two years. High adult survival of long-lived species made the joint model with recovery data ineffective at reducing terminal bias because of small-sample bias. The naive constraint model (last and penultimate temporary emigration parameters made equal), was the least efficient, although still able to reduce terminal bias when compared to an unconstrained model. Joint analysis of several sources of data improved parameter estimates and reduced terminal bias. Efforts to incorporate or acquire such data should be considered by researchers and wildlife managers, especially in the years leading up to status assessments of species of interest. Simulation modeling is a very cost-effective method to explore the potential impacts of using different sources of data to produce high-quality demographic data to inform management.

The Role of Publication Selection Bias in Estimates of the Value of a Statistical Life

Meta-regression estimates of the value of a statistical life (VSL) controlling for publication selection bias often yield bias-corrected estimates of VSL that are substantially below the mean VSL e...

Revisiting the Effect of Capture Heterogeneity on Survival Estimates in Capture-Mark-Recapture Studies: Does It Matter?

Recently developed capture-mark-recapture methods allow us to account for capture heterogeneity among individuals in the form of discrete mixtures and continuous individual random effects. In this article, we used simulations and two case studies to evaluate the effectiveness of continuously distributed individual random effects at removing potential bias due to capture heterogeneity, and to evaluate in what situation the added complexity of these models is justified. Simulations and case studies showed that ignoring individual capture heterogeneity generally led to a small negative bias in survival estimates and that individual random effects effectively removed this bias. As expected, accounting for capture heterogeneity also led to slightly less precise survival estimates. Our case studies also showed that accounting for capture heterogeneity increased in importance towards the end of study. Though ignoring capture heterogeneity led to a small bias in survival estimates, such bias may greatly impact management decisions. We advocate reducing potential heterogeneity at the sampling design stage. Where this is insufficient, we recommend modelling individual capture heterogeneity in situations such as when a large proportion of the individuals has a low detection probability (e.g. in the presence of floaters) and situations where the most recent survival estimates are of great interest (e.g. in applied conservation).

Growth, survival, and tag retention of steelhead trout (Oncorhynchus mykiss) and its application to survival estimates

Steelhead trout, Oncorhynchus mykiss, are known to expel acoustic tags which can negatively bias population survival estimates. Tag burden may also affect the development and behavior of smolts, thereby impacting the results of population and behavioral studies. We monitored the growth, condition, and tag expulsion rate of steelhead trout in similar-sized individuals and used these data to adjust survival rates from an acoustic telemetry study conducted in the Sacramento River. Eighty fish were surgically implanted with tags—40 with cylindrical tags of 9 mm diameter and 21 or 24 mm length (V9, Vemco Ltd) and 40 with a 7 mm diameter and 20 mm length tag (V7, Vemco Ltd)—to examine the impact of tag size on peritoneal retention and survival rate of juvenile steelhead trout. A total of 20 % (16/80) of all tags were expelled by smolts during the 143-day study. Ten V9 tags were expelled between day 18 and day 66. Six V7 tags were expelled between day 21 and day 143. A statistical difference was found for retention rate by surgeon even though the surgeons were of equal experience and received the same training. There were no significant differences in the tag retention rate in relation to the tag/body weight ratio, or in growth (weight or fork length) among the control, V7 or V9 treatment groups over the duration of the study. All individuals survived throughout the experiment. Two methods were used to adjust the survival estimates of an acoustic telemetry data set from the Sacramento River based on the tag retention study. First, a simple individual censorship approach in Program MARK was utilized and next ATLAS, a software program designed to compensate for bias in survival estimates caused by tag failures was used. The results of the adjusted survival estimates were not significantly different from the unadjusted rates suggesting that it may be more important to focus on improving surgical techniques to reduce tag expulsion rather than adjusting survival estimates dependent on the study. The surgical techniques utilized in this study did not have significant impacts on the growth rates of either of the tag treatment groups compared to the control. However, tag retention was an issue regardless of the size and weight of the implanted tag and the size of the steelhead.

Conditional modelling of ring‐recovery data

Summary1. Ring‐recovery data can be used to obtain estimates of survival probability which is a key demographic parameter of interest for wild animal populations. Conditional modelling of ring‐recovery data is needed when cohort numbers are unavailable or unreliable. It is often necessary to include in such analysis a recovery probability that is declining as a function of time, and failure to do this can result in biased estimates of annual survival.2. Corresponding estimates of survival probability need to be reliable in order for correct conclusions to be drawn regarding the effects of climate change.3. We show that standard logistic modelling of a decline in recovery probability is unsatisfactory, and propose and investigate a range of alternative procedures.4. Methods are illustrated by application to a recovery data set on grey herons. The model selected is a scaled‐logistic model, and it is shown to provide a unifying analysis of several data sets collected on different common bird species. The model makes specific predictions, providing potential new insights and avenues for ecological research. The wider performance of this model is evaluated through simulation.5. In this study, we propose a new scaled‐logistic model for the analysis of ring‐recovery data without cohort numbers, which incorporates a reporting probability that declines over time. The model is shown to perform well in simulation studies and for both a single real data set and several real data sets in combination. Its use has the potential to reduce bias in estimates of wild animal survival that currently do not incorporate such reporting probabilities. Alternative models are shown to possess undesirable features.

PaFLO: Pazopanib with 5-fluorouracil, leucovorin, and oxaliplatin (FLO) as first-line treatment in advanced gastric cancer: A randomized phase II study of the Arbeitsgemeinschaft internistische Onkologie (AIO).

TPS4138 Background: VEGF inhibition in gastric cancer shows promising improvement of remission rate and progression-free survival (Ohtsu et al., JCO 2011). Pazopanib is an orally available tyrosine kinase inhibitor (TKI) selectively inhibiting VEGFR-1, -2, -3, c-kit and PDGFR. It is approved for treating renal cell cancer. A phase-I trial showed good tolerability of pazopanib with full-dose FOLFOX in solid tumors (Brady et al., ASCO, 2009). FLO is a widely used combination for advanced gastric cancer recommended in national guidelines. Methods: 75 Patients with HER-2-negative locally advanced or metastatic adenocarcinoma of the stomach or the gastro-esophageal junction will be randomized in a 2:1 ratio to A: FLO (F 2600mg/m2 as 24h infusion, L 200mg/m2, O 85 mg/m2) d1 + pazopanib (800mg) d1-14 and B: FLO; repeated for 12 2-week cycles, followed by a maintenance therapy with pazopanib alone in A and an observation period in B until disease progression. Primary endpoint is progression-free survival rate (PFSR) at 6 months, secondary endpoints are PFSR at 9 and 12 months, median PFS, response rate, duration of response, toxicity, tolerability and overall survival. Additionally, we evaluate the predictive and prognostic relevance of PIGF, VEGF, and the respective soluble receptors sVEGFR1 and sVEGFR2 as biomarkers for clinicopathological parameters, clinical response to treatment and tumor volume change. Based on a phase-III trial demonstrating a 6-month PFSR of 44% with FLO (Al-Batran et al., 2008), we estimate a 6-month PFSR of 55% in the experimental group. Given an alpha error of 0.1 and a beta error of 0.2 in a Simon 2-stage minimax design, in the first stage ≥12 of 30 patients need to be progression free at 6 months to continue and after the second stage ≥25 of 50 patients should be progression free at 6 months to justify further evaluation. Randomization is performed to estimate selection bias according to pazopanib-specific exclusion criteria for comparison with historical data. Study protocol received ethics committee approval in November 2011 and is currently recruiting patients in 15 AIO centers.

Comparison of Methods to Correct Survival Estimates and Survival Regression Analysis on a Large HIV African Cohort

ObjectiveThe evaluation of HIV treatment programs is generally based on an estimation of survival among patients receiving antiretroviral treatment (ART). In large HIV programs, loss to follow-up (LFU) rates remain high despite active patient tracing, which is likely to bias survival estimates and survival regression analyses.MethodsWe compared uncorrected survival estimates derived from routine program data with estimates obtained by applying six correction methods that use updated outcome data by a field survey targeting LFU patients in a rural HIV program in Malawi. These methods were based on double-sampling and differed according to the weights given to survival estimates in LFU and non-LFU subpopulations. We then proposed a correction of the survival regression analysis.ResultsAmong 6,727 HIV-infected adults receiving ART, 9% were LFU after one year. The uncorrected survival estimates from routine data were 91% in women and 84% in men. According to increasing sophistication of the correction methods, the corrected survival estimates ranged from 89% to 85% in women and 82% to 77% in men. The estimates derived from uncorrected regression analyses were highly biased for initial tuberculosis mortality ratios (RR; 95% CI: 1.07; 0.76–1.50 vs. 2.06 to 2.28 with different correction weights), Kaposi sarcoma diagnosis (2.11; 1.61–2.76 vs. 2.64 to 3.9), and year of ART initiation (1.40; 1.17–1.66 vs. 1.29 to 1.34).ConclusionsIn HIV programs with high LFU rates, the use of correction methods based on non-exhaustive double-sampling data are necessary to minimise the bias in survival estimates and survival regressions.

Estimating survival of a streamside salamander: importance of temporary emigration, capture response, and location

Estimating survival for highly secretive aquatic animals, such as stream salamanders, presents numerous challenges. Salamanders often spend a considerable time in refugia where they are difficult to capture. Few studies have calculated vital rates for stream salamanders, yet the need is substantial as they are threatened by a wide range of land-use stressors, especially urban development. In this study, we used 34 months of continuous field samples collected at an urban and undisturbed stream and robust design mark-recapture analysis to evaluate the importance of temporary emigration, capture response, and location on survival estimates of the salamander Desmognathus fuscus. We constructed a set of candidate models incorporating combinations of time- and location-varying capture and recapture probabilities, capture responses, temporary emigration, and survival estimates and ranked models using Akaike’s Information Criterion. We found strong support for month-specific capture probabilities, recapture probabilities, temporary emigration and a negative behavioral response to capture in the majority of months. We found no support for variation in capture probabilities, recapture probabilities, and temporary emigration between locations. However, we found that location strongly influenced survival estimates. Specifically, survival estimates were significantly higher at the undisturbed site than at the urban site. Our results emphasize the importance of estimating capture probabilities, recapture probabilities, capture response, and temporary emigration when evaluating survival in highly secretive aquatic animals. Failure to account for these population parameters will likely yield biased estimates of survival in freshwater animal populations.