Hemorrhoidal disease is one of the most prevalent diseases worldwide, with a global incidence of 4.4% to 36%, depending on the population studied and the diagnostic criteria employed.1,2 It has a considerable effect on patient quality of life due to symptoms of rectal bleeding, pruritus, pain, and prolapse.3 Even though most cases can be managed with conservative treatment, a considerable number require surgical intervention. Hemorrhoidectomy is the treatment of choice in advanced or refractory cases.4,5

The development of hemorrhoidal disease has been associated with different risk factors, including chronic constipation, excessive straining, pregnancy, aging, sedentary lifestyle, and elevated body mass index (BMI).6,7 However, there are discrepancies in the literature as to the true impact of those factors and their predictive value for determining the need for surgery.8,9

Given the socioeconomic burden of the disease and the need for optimizing its management, it is crucial to more accurately identify the risk factors associated with progression to hemorrhoidectomy.10,11 Thus, the present study aimed to provide evidence on the clinical variables that could be associated with symptom duration in patients ultimately requiring surgery.

A retrospective, cross-sectional, observational study was conducted at the Hospital Universitario de Puebla. It included 56 patients diagnosed with grade III or IV hemorrhoidal disease, according to the Goligher classification, all of whom underwent hemorrhoidectomy between 2019 and 2023. The study design followed the STROBE checklist for observational research. The aim was to examine the clinical and demographic factors associated with symptom duration before surgical intervention, without causal inference or predictive modeling.

The independent variables were sex, age, age at symptom onset, smoking status, constipation, and BMI. All patients had received conservative treatment with diosmin/hesperidin and dietary fiber for at least 6-8 weeks, with unsatisfactory clinical response. None of the patients had been treated with previous outpatient procedures, such as elastic band ligation or sclerotherapy and no clinical or functional diagnosis of pelvic floor dysfunction was documented.

A descriptive analysis of the qualitative variables (frequencies and percentages) and quantitative variables (means and standard deviations) was performed. Distribution normality was evaluated through skewness, kurtosis, and the Shapiro-Francia test. The chi-square test was used to compare the qualitative variables and the one-way analysis of variance (ANOVA) for the quantitative variables.

Models of linear regression were constructed for identifying associations between the clinical variables and symptom duration. Three primary models were evaluated: (1) symptom duration as a function of sex and age, (2) symptom duration as a function of age at onset, and (3) an interaction model between sex and age at onset. Model robustness was tested through K-Fold cross-validation (K = 5), balancing accuracy and error stability.

Because of the retrospective and exploratory nature of the study, an a priori inferential calculation of the sample size was not carried out. However, the number of patients included in the study was based on the methodological criterion of at least 10 observations per independent variable in the multivariable model. Considering five main predictors, the minimum sample size recommended was 50 subjects, which was surpassed (n = 56). Therefore, the sample was considered adequate for exploratory analyses without a high overfitting risk. Nevertheless, a post hoc power analysis was carried out, based on the most significant regression coefficient between age at onset and symptom duration (β = –0.31), obtaining 62.5% statistical power with a level of significance of α = 0.05 and a sample size of 56 patients, suggesting limited power for detecting effects of that size and a non-negligible possibility of a type II error, signifying that the results be cautiously interpreted.

The present study was conducted respecting international research and ethics principles for health research and conforming to current ethical standards. It is a retrospective, observational study with previously collected anonymized data, in which no experiments on humans or animals were carried out. Data that could identify patients were not included, guaranteeing their confidentiality. In accordance with the ethical guidelines of the Hospital Universitario de Puebla, approval from the ethics committee was not required for the analysis and publication of the routinely collected clinical data. Due to the retrospective nature of the study, informed consent from the patients was not necessary.

Of the 56 patients included in the study, 53% were men and 47% were women. A total of 56% did not smoke and 64% presented with constipation. The mean BMI was 27 kg/m2, and the mean time from symptom onset to surgery was 6.7 years.

• •

  Years of disease vs sex and age: The regression analysis showed that sex was significantly related to years of disease (p = 0.0427); men had a mean of 4.89 more years of disease than women (β = –4.89, 95% CI –9.61 to –0.1). Age showed a non-significant positive association with disease duration (β = 0.16 years per additional year of age; 95% CI –0.03 to 0.35; p = 0.0907) Neither statistical significance nor clinical relevance were reached in this model (Fig. 1)

  
  

  Figure 1.

  Boxplot comparing years of disease by sex.
• •

  Years of disease vs age at onset: The linear regression analysis showed a significant inverse relation (p < 0.001): for each additional year in age at onset, disease duration decreased by approximately 0.31 years (β = –0.31, 95% CI –0.46 to –0.15) (Fig. 2).

  
  

  Figure 2.

  Scatterplot showing the relation between age and years of disease, with linear regression.
• •

  Interaction between sex and age at onset: The interaction between sex and age at onset, in relation to years of disease, was significant (p < 0.001), indicating that the effect of age at onset on years of disease varied by sex, with a significant interaction (β = 0.53; 95% CI 0.24 to 0.82; p < 0.001). In men, each additional year in age at onset corresponded to a smaller decrease in years of disease, compared with women (β = 0.53, 95% CI 0.24 to 0.82) (Fig. 3).

  
  

  Figure 3.

  Scatterplot showing the interaction between age at onset and sex, with confidence intervals.

The model showed a mean squared error (MSE) of 55.66, with a standard deviation of 25.17, suggesting moderate variability in its performance.

The present study identified significant associations between sex (β = –4.89; 95% CI –9.61 to –0.17; p = 0.0427) and age at symptom onset (β = –0.31; 95% CI –0.46 to –0.15; p < 0.001) with symptom duration prior to hemorrhoidectomy. Men had longer disease duration than women, and earlier age at onset was associated with longer symptom duration.

Said results concur with those of studies reporting that men tend to delay seeking medical care, possibly due to differences in pain perception and self-care behaviors.10 The interaction between sex and age at onset has also been described in other settings, in which women have been found to seek medical attention earlier, which could explain the shorter symptom duration before surgery.1,12–14

Conversely, constipation, traditionally considered an important risk factor for hemorrhoidal disease,1,2 showed no significant association with symptom duration in our analysis, contrasting with previous reports linking chronic constipation to greater hemorrhoidal severity. This suggests that other factors, such as pelvic floor dysfunction or bowel habits, could influence disease progression, but they were not assessed in the present cohort.

Similarly, BMI showed no significant relation to disease duration (β –0.087; 95% CI –0.69 to 0.52; p > 0.05), consistent with studies that analyze its association with hemorrhoidal severity.9,15,16 However, other authors have described conflicting results, indicating the need for studies that incorporate more detailed body composition and lifestyle measures for clarifying said association.

Lastly, our findings align with the management guidelines of the American Society of Colon and Rectal Surgeons (ASCRS) and the European Society of Coloproctology (ESCP), which recommend hemorrhoidectomy in patients with grade III or IV disease and persistent symptoms despite conservative treatment.17,18

We chose a linear regression model due to the absence of a non-surgical comparison group. In an ideal design that included individuals successfully treated conservatively, logistic regression could have been used to model the likelihood of surgical indication. Our analysis was limited to an exclusively surgical cohort, focusing on symptom duration rather than surgical decision as the outcome, thus reducing generalizability and predictive applicability of the model.

Likewise, the present study did not evaluate pelvic floor function parameters or bowel habits through objective tests or validated questionnaires, limiting the interpretation of the lack of association between constipation and clinical progression. In addition, the retrospective, cross-sectional design precludes causal inference and introduces potential selection and reporting biases.

The reduced sample size (n = 56) also limited statistical power and the ability to detect more subtle associations and hampered establishing clinically useful cutoff points in the exploratory model. This was reflected in the evaluation of an alternative logistic regression model for symptoms lasting ≥ 2 years (area under the curve [AUC] 0.77), in which no cutoff point achieved an adequate balance between sensitivity and specificity.

Future prospective, multicenter studies that include non-surgical patients and collect data on pelvic floor function, quality of life, and postoperative outcomes (pain, recurrence, satisfaction), will enable the development of more robust multivariable models. The integration of advanced statistical techniques, such as LASSO regression or machine learning algorithms, could improve the accuracy and clinical applicability of predictive scores or nomograms, incorporating key variables, such as sex and age at onset, which showed significant association in our study.

The present study identified sex and age at disease onset as factors associated with symptom duration in patients that required hemorrhoidectomy. Even though factors, such as constipation and BMI, have traditionally been considered relevant, they showed no significant association in our analysis. Our findings underscore the importance of considering complex interactions between clinical and demographic variables for predicting hemorrhoidal disease progression.

However, because of its retrospective and cross-sectional design, the results should be interpreted as exploratory and cannot establish causal relationships or support predictive inferences with generalizable clinical validity.

Several limitations should be considered when interpreting the results of the present study. First, the small sample size (n = 56) limits the general statistical power of the analysis, increases the risk of type II error, and restricts confidence in negative findings. In addition, the retrospective and cross-sectional design introduces potential selection and data collection biases, precluding causal inference.

Even though the cohort was homogeneous regarding disease grade (grade III or IV hemorrhoids) and previous medical treatment, detailed information on other functional comorbidities, such as anorectal dysfunctions confirmed through physiologic tests, were not included. Likewise, other potentially influential factors were not evaluated, such as dietary habits, physical activity, subjective response to conservative treatment, or institution-specific criteria for surgical indication.

The performance of the predictive model was also limited, as indicated by the relatively high MSE (55.66 ± 25.17), reinforcing the need for more robust multivariable approaches. Future studies could benefit from advanced statistical techniques, such as LASSO regression or machine learning algorithms (random forest), along with external validation.

Despite these limitations, our study provides initial evidence on significant associations between demographic variables (age, sex) and symptom duration in patients with hemorrhoidal disease who require surgery. To improve the generalizability and clinical usefulness of our findings, we suggest conducting larger prospective, multicenter studies that include data on pelvic floor function, quality of life, postoperative outcomes, and the development of predictive tools applicable in clinical practice.

No financial support was received in relation to this study.